Turkish m-reduplication: a case of simulative plural

  1  

Turkish m-reduplication: a case of similative plural 1. Introduction

In Turkish colloquial speech one frequently encounters m-reduplication1: (1) a. Bulut m-ulut topla-dı-Ø, şimdi yağ-acak-Ø. cloud m-cloud gather-PST-3SG now rain-FUT-3SG ‘Clouds and the like gathered, it will rain soon.’ b. Güzel m-üzel bir kız-Ø ama biraz çekin-gen-Ø. pretty m-pretty one girl-NOM but a_little shy-NMLZ-3SG ‘She is pretty and the like but she is shy.’ c. Mektuplar-ı yaz-dı-Ø m-az-dı-Ø ama sonra unut-tu-Ø. letter-ACC write-PST-3SG m-write-PST-3SG but then forget-PST-3SG ‘S/he wrote letters and the like but then s/he forgot them.’ d. Jonathan kurabiye-yi bütün m-ütün ye-di-Ø. Jonathan cookie-ACC all m-all eat-PST-3SG ‘Jonathan ate the cookies whole, like without chewing.’ Although this type of reduplication has been fleetingly mentioned in Turkish grammars (Göksel

& Kerslake 2005:99-100, Ketrez 2012:261, Underhill 1976:437-438, Hatipoğlu 1981, among

others), it has neither been explored in depth nor given any explanation2.

In this study, we discuss fieldwork data in detail, and then propose an account. We argue

that m-reduplication is a case of lexical number, namely similative plural, that applies across

categories.

The paper is structured as follows. Section 2 summarizes what we know about m-

reduplication based on different sources, and identifies the questions we raise. Section 3 presents

fieldwork-driven generalizations on the morphosyntactic distribution as well as semantic and

stylistic restrictions on m-reduplication. Section 4 defines tools necessary for an analysis, i.e. it

spells out our theoretical assumptions, and outlines our proposal. Section 5 discusses our

conclusions, raising further questions about the Turkish language as well as cross-linguistic

questions.

                                                                                                               1 All data are from our own fieldwork unless indicated otherwise. 2 Müller’s (2004) in depth study of Turkish reduplicative patterns is an exception. With respect to m-reduplication, he is interested in the historical origins of m-reduplication and how/whether it has spread (for discussion, see Müller 2004:320-329). Our interest lies in the synchronic use and an account of it. In that sense, there is little overlap with Müller (2004). The nature of the data is different, too. Müller’s (2004) data are mainly lists of words, without a clausal or discourse context. We discuss the syntactic and discourse context of m-reduplicated expressions.

  2  

2. What we do and do not know about m-reduplication

This section sums up what we know about m-reduplication in Turkish, and identifies

what still needs to be established about it.

Phonologically, m-reduplication is sensitive to the structure of the base (Underhill

1976:437, Göksel & Kerslake 2005:99). If the base starts with a consonant, the reduplicated

entry is repeated, deleting the consonant, which is replaced by the prefix m-:

(2) kapı-Ø m-apı-Ø door-NOM m-door-NOM ‘doors and the like’ Göksel & Kerslake 2005:99 However, this pattern of reduplication is prohibited if the base starts with an m- (Underhill

1976:437). If the base starts with a vowel, m- is just prefixed on the base:

(3) etek-Ø m-etek-Ø skirt-NOM m-skirt-NOM ‘skirts and the like’ Göksel & Kerslake 2005:99

Morphologically, this structure would be considered marked: Turkish hardly uses

prefixes (Göksel & Kerslake 2005:67), and yet this particular pattern of reduplication of the base

that includes prefixation is highly productive (Underhill 1976:438, own fieldwork). Also Turkish

words do not begin with m- (Hatipoğlu 1981)

Another morphologically notable fact is that this kind of reduplication is attested in

phrasal compounds. As illustrated below, the entire compound is repeated (indicated by brackets),

but only the first element gets the prefix m-:

(4) Ben adam-Ø [tarih-Ø hoca-sı-ymış] [m-arih-Ø… I man history-NOM teacher-ACC-EV/PF m-history-NOM … hoca-sı-ymış] anla-ma-m Fransız-Ø tarih-i-ni o-ndan teacher-ACC-EV/PF care-NEG-1SG French-NOM history-ACC-GEN 3SG-ABL … daha… iyi bil-iyor-um. more better know-PROG-1SG ‘I don’t care if he is a history teacher or whatever. I know more about French history than he does.’ Göksel & Kerslake 2005:100

  3  

We consider this significant in that m-reduplication applies beyond word level.

With respect to what exactly m-reduplication means, there is some slight variation in the

sources. Underhill (1976:437) transmits it as English ‘x and stuff’, similar to Müller (2004).

Göksel & Kerslake (2005:99-100 ) oscillate between ‘x and the like’, ‘x and whatever’ and ‘etc.’.

Ketrez (2012:261 ) conveys it as ‘etc.’ All sources agree on this: m-reduplication is colloquial

and not attested either in formal or written Turkish. Assuming that linguistics concerns itself

primarily with oral language in all its guises, we consider this an unfortunate omission in the

discourse on Turkish linguistics.

This concludes what we know about m-reduplication. There are some further

observations, but they are inconclusive due the lack of data, generalizations, and analysis. For

example, all three sources previously cited mention that m-reduplication is observed in different

categories, yet specific data examples are lacking. However, neither is the inventory of the

categories provided nor are restrictions on usage discussed. And, to the best of our knowledge,

there has been no attempt to provide an explanation for this pattern of reduplication. In this study,

we address the following questions:

(i) What is the distribution of m-reduplication?

(ii) What are the constraints on the distribution of m-reduplication?

(iii) How do we account for m-reduplication?

3.Data: generalizations on m-reduplication In this section, we present fieldwork data on m-reduplication. First, we walk through each

category where m-reduplication is attested (3.1). The findings are summarized in Table 1.

Table 1. M-reduplication across categories Noun Pronoun Adj/Adv Numeral Quantifier Verb Complementizer m-reduplication

✓

✓

✓

✓

✓

✓

n.a. a shift in category

Next, we discuss the plasticity of m-reduplicated expressions, i.e. what they mean (3.2).

Then we address the restrictions on the distribution of m-reduplication (3.3). We find that

restrictions are semantic, contextual, and stylistic.

  4  

Finally, we summarize the findings and raise the questions necessary for providing an

explanation (3.4).

3.1.Distribution of m-reduplication Nouns. M-reduplication is most common in nouns. It expands the meaning of the base noun to

include other nouns, related to the base noun. In the case of nouns, the reduplicated noun

introduces the referent for the reduplicated string:

(5) a. Kitap-Ø m-itap-Ø oku-mak iyi gel-ir.

book-NOM m-book-NOM read-NMLZ good come-AOR.3SG ‘Reading book and the like makes you feel better.’

b. Elma-Ø m-elma-Ø dolap-ta dur-uyor. apple-NOM m-apple-NOM fridge-LOC stay-PROG.3SG ‘Apples and the like stay in the fridge.’ In (5), either the plural or the singular interpretation of the base noun is available. This is due to

the general properties of Turkish bare nouns. For bare nouns, an overt plural marking is optional.

If not marked for plural, Turkish bare nouns may be interpreted as either singular or plural (for a

recent in depth account of Turkish nominal number, see Görgülu 2012). However, once the

plural is marked, the singular interpretation is ruled out, in both simplex as well as the

reduplicated forms:

(7) a. kitap-Ø b. kitap- Ø-lar book-NOM book.NOM-PL ‘book’ or ‘books’ ‘books’ c. kitap-Ø-lar m-itap-Ø-lar d. *kitap-Ø m-itap-Ø-lar book-NOM-PL m-book-NOM-PL book-NOM m-book-NOM-PL ‘books and the like’ ‘book and the like’ Inflectional suffixes, when present, are required on both elements: the base stem and the

reduplicated stem (8a). If only one of the two elements carries the inflectional suffix, the string

becomes ungrammatical (8b-c).

  5  

(8) a. kitap-ta m-itap-ta book-LOC m-book-LOC ‘in the books and the like’

b. *kitap-ta m-itap book- LOC m-book c. *kitap-Ø m-itap-ta book-NOM m-book-LOC The same is true of derivational affixes: they, too, when present, are required on both elements,

or else the derivation becomes ungrammatical3:

(9) a. kitap-çı m-itap-çı book-NMLZ m-book-NMLZ ‘bookseller and the like’ b. *kitap-çı m-itap-Ø book- NMLZ m-book-NOM c. *kitap-Ø m-itap-çı book-NOM m-book-NMLZ Pronouns. M-reduplication is also attested in pronouns: (10) a. Ben m-en yok! I m-I not_exist Literally: ‘There is no me/my life’ Pragmatically: ‘You should accept it!’ b. O- Ø m-o- Ø bura-da otur-mu-yor-∅. s/he-NOM m-s/he-NOM here-LOC live-NEG-PROG-3SG ‘S/he and anyone/anything related to her/him do not live here.’ Pronominal m-reduplication is particularly open-ended with respect to what exactly it means, if

compared to nouns. In the case of nouns, the base noun introduces the referent for the

reduplicated string, while the reduplicated noun introduces entities related to that noun. Thus, the

lexical meaning of that base noun determines the referential range of the reduplicative string (see

section 3.2 for more discussion). Pronouns do not have an inherent referent; rather, that referent

                                                                                                               3 The facts about the behavior of inflectional and derivational suffixes are not specific to nominals, but hold across other categories. Once affixation introduced, it is obligatory on both elements whenever m-reduplication applies. In the interests of space, we will not redundantly repeat these tests on other categories in the remainder of the paper.

  6  

is picked up in discourse. Consequently, the referential possibilities of a reduplicated pronoun

string are decided by discourse, and therefore potentially infinite.

Adjectives and adverbs. Both adjectives (11a,b) and adverbs (11c) can be m-reduplicated, too.

The m-reduplicated string denotes a particular property plus other properties that could be related

to the property denoted by the base:

(11) a. Akıllı m-akıllı ama biraz tembel-∅. smart m-smart but a_little lazy-NOM. ‘S/he is smart and the like but a little bit lazy at the same time.’

b. Güzel m-üzel ama biraz bencil-∅. pretty m-pretty but a_little selfish-NOM ‘S/he is pretty and the like but s/he is a little bit selfish.’

c. Tahminen m-ahminen birşey diye-me-m. approximate m-approximate something say-NEG-1SG ‘I don’t say anything approximate.’

Numerals. Numerals can be m-reduplicated as well. M-reduplication has an effect of an increase

in number, albeit with vagueness as to what the increased number is. That is, the number is not

exact, and is always bigger than the base numeral. In (12a), it means that while it is unclear

exactly how many guests came, it is certain that more than twenty came, and not less than twenty.

In (12b), it means that while it is unclear exactly how much alcohol the person consumed, it is

certain that s/he consumed more than two glasses:

(12) a. Yirmi m-irmi misafir-∅ gel-di-∅. Twenty m-twenty guest-NOM come-PST-3SG. ‘Around twenty guests came.’ b. Iki m-iki bardak-∅ içki-∅ iç-ti-∅. Two m-two glass-NOM alcohol-NOM drink-PST-3SG. ‘S/he drank around two glasses of alcohol.’ Quantifiers. M-reduplication applies to quantifiers, too. A particular meaning depends on the

lexical content of a particular quantifier:

(13) a. Bütün m-ütün yut-ma- Ø yemeğ-i! Whole m-whole swallow-NEG-IMP meal-ACC ‘Don’t swallow the meal in a gulp and the like.’

  7  

b. Birkaç m-irkaç kişi-∅ gel-ecek-∅ . Some m-some people-NOM come-FUT-3SG ‘Some people (more/less) will come.’ Verbs. Finally, m-reduplication is attested on verbs. It has an effect of including additional

events (or states) to the event (or state) denoted by the base verb.

(14) a. Çikolata-yı ye-di-Ø m-e-di- Ø sonra uyu-du- Ø . chocolate-ACC eat-PST-3SG m-eat-PST-3SG then sleep-PST-3SG ‘S/he ate or just took a little bit of the chocolate and then slept.’

b. Malia-∅ kitab-ı oku-du- Ø m-oku-du Ø… Malia-NOM book-ACC read-PST-3SG m-read-PST-3SG

… ama yarın sınav-da zorlan-acak- Ø . but tomorrow exam-LOC sweat-FUT-3SG ‘Malia read m-read(skimmed or scanned) the book but she will have hard times at the exam tomorrow.’

Clause level functional morphemes. M-reduplication may also be found on complementizers,

such as, e.g. conjunction ama ‘but’:

(15) Ama m-ama yok! but m-but not_exist ‘No buts!’ Context: Hasan wants to go out, and he wants to take his friend along. Hasan’s friend does not have any money to go out, so he puts several excuses (But I do not have the money! But I cannot borrow so much from you!), trying to get out of the awkward situation. Hasan would not hear any of this, and utters (15). However, m-reduplication in this case is restricted to an instance where the complementizer is

used as a noun, as the context makes clear, i.e. ‘No buts!’ could be equated with ‘No excuses!’

here. Thus, the complementizer undergoes a shift in category and is classified as a noun rather

than a functional morpheme. If one were to use the m-reduplicated ama ‘but’ in its proper

function of complementizer in a sentence the result would be ungrammatical.

(16) *Ayşe-∅ gel-me-yecek ama m-ama gel-ebil-ir.

Ayşe-NOM come-NEG-FUT but m-but come-PSB-AOR.1SG

  8  

Therefore we conclude that m-reduplication does not apply to functional categories at clause

level.

3.2. M-reduplicated entries: fuzziness in meaning M-reduplication is interpreted as introducing a sort of vagueness, e.g. ‘x and the like’

across categories. But what exactly does this mean? What is the range of ‘…and the like’ that

can be associated with the base? Our fieldwork data show that two factors determine the range

of ‘…and the like’: the lexical meaning of base and the context of the utterance. We illustrate

both generalizations in detail.

The lexical meaning of the base determines the range of ‘…and the like’ in that it

delimits what the m-reduplicated phrase can mean without any context, inherently. The more

specific the lexical item is, the more delimited the range of the ‘…and the like’.

For example, a nominal expression like öğrenciler möğrenciler ‘student and the like’

may include any individuals pertaining to learning environment implied by the lexical meaning

of ‘student’:

(15) Öğrenci-∅-ler m-öğrenci-∅-ler toplan-dı- Ø . student-NOM-PL m-student-NOM-PL gather-PST-3PL ‘Students and such gathered.’ Interpretation: Students and other staff (e.g. teachers, teaching assistants, graders, other people who are in some way related to teaching/studying…)

Similarly, an adjectival expression like beyaz meyaz ‘white and such’ may extend to colors

closely associated with white color:

(16) Ev-i beyaz m-eyaz boya-dı-m. house-ACC white m-white paint-PST-1SG ‘I painted the apartment white and the like’. Interpretation: White and other shades of whitish colors, maybe eggshell or a shade of grayish

white, but nothing like red or blue or orange.

  9  

3.3. Restrictions on m-reduplication

It is important to note that m-reduplication is optional. We have yet to discover an

environment where m-reduplication would be obligatory. However, we did discover some

restrictions on its use. In this section we address the restrictions. The restrictions are: (i)

phonological; (ii) semantic; (iii) contextual; (iv) and stylistic.

Phonological restrictions. The phonological conditions for the use of m-reduplication have been

noted in the literature, as reported in Section 2, therefore we will not return to this question.

Semantic restrictions. M-reduplication is sensitive to the superordinate category of the base. The

m-reduplicated construction has some flexibility as to what it means as long as the interpretation

is within the limits of the superordinate category, e.g. a human in the case of (17).

(17) Ayşe-∅-ler m-ayşe-∅-ler aksam yemegi-ne gel-iyor-lar. Ayşe-NOM-PL m-ayşe-NOM-PL night meal-DAT come-PROG-PL ‘Ayşe mayşe are coming for dinner!’ Context A: acceptable

Celia and Ayşe are close friend since childhood. Basically, they often come to each other’s

houses for dinner, and nobody minds if they bring friends.

In the scenario A, (17) can refer to any people coming along for dinner with Ayşe , whether these

people are Ayşe’s friends or mutual friends of Ayşe and Cecilia.

Context B: unacceptable

Ayşe is friends with Celia, but under this scenario, Ayşe is the only friend of Celia and Celia is

the only friend of Ayşe. However, Ayşe has five pet dogs and she sometimes brings either all the

dogs or some of the dogs over for dinner.

Even though the dogs are Ayşe ’s and they are animate beings, the utterance in (17) cannot refer

to scenario in B, because, as speaker consultants put it, ‘dogs are not human.’

Context C: unacceptable

Under this scenario, Ayşe is still friends with Celia, but Ayşe has no other friends, or dogs or

family. She lost them all in a terrible car accident. She is gone quite mad after the accident:

mutters to herself, collects pebbles, and calls these pebbles friends.

  10  

Even though Celias’s family are trying to help Ayşe, and accommodate her as much as it is

possible, the utterance in (17) cannot refer to the scenario in C, because the metaphorical

extension does not make pebbles a subtype of humans.  

Contextual restrictions. At times, a particular interpretation of an m-reduplicated expression

can be attained by manipulating the context of an utterance (as long as the semantic

supertype~type conditions are not violated). For example, ‘drink’ in (18) can refer to any of the

following:

(18) Hadi önce iç-elim m-iç-elim, sonra konuş-ur-uz.

let’s first drink-COND m-drink-COND then talk-AOR-1PL ‘First let’s have a drink m-drink, and then we talk!’  

Context A

(18) is uttered in the evening, in a bar which serves exclusively whisky. You are hanging out

with friends who love whisky, and this is a party to celebrate a new member of the group joining

the weekly whisky tastings. In this case, (18) can refer just to whisky, not just any drink.

Context B

(18) is uttered in a break of runners who are training for a marathon. They have just completed a

20 km, and have a 5 minute drinking break before they ran the remaining 22 km.

In this case, (18) can refer just to water and sports drinks, not just any drink.

Context C

(18) is uttered in tea shop where people go to drink tea. No other drink is served.

In this case, (18) can refer just to tea, not just any drink.

Context D

(18) is uttered after a long day at work, people had to meet a deadline and had no time to drink or

eat.

In this case, (18) can refer to any drink and maybe also some foods. Plus, it can be alcoholic

drinks, too. In Turkey, if you pass a long day at work and you say this to your friends, it means

that you want to eat something (maybe some appetizers or soup) and drink mostly alcoholic

beverages (probably one or two bottles of Turkish beer).

Context E

(18) is uttered in a mixed company of vampires and humans. Thus, its meaning now also

potentially includes ‘blood’, given the tastes of vampires.  

  11  

Stylistic restrictions. M-reduplication does not surface in formal discourse, while it is perfectly

acceptable to use in colloquial speech. For example, (19a-b) are both perfectly grammatical and

convey similar meaning, yet their stylistic restrictions differ.

(19) a. Adam-∅ suç-∅ m-uç-∅ işle-miş-∅. man-NOM crime –NOM m-crime-NOM commit-REP.PST-3SG ‘The man committed crime and the like (maybe a range of different crimes such as burglary, assassination, cyber crime).’ b. Bu kişi-∅ hırsızlık-∅, cinayet-∅, bilişsel suç-∅… this person-NOM burglary-NOM assassination- NOM cyber_crime-NOM …işle-miş-tir. commit-REP-PST-3SG ‘This person committed to burglary, assassination, cyber crime.’ (19a) could be uttered in an informal context, by, e.g. a lawyer chatting with a friend in a pub.

However, (19a) would not be acceptable to utter in a court of law, while the lawyer does his job

in a highly formal context. It is (19b) that matches the required formality of the court.

Conversely, it may be slightly odd to use (19b) in the context of an informal pub chat.

Lastly, there is one more specifically semantic effect of m-reduplication: it induces

pejorative speaker perspective on whatever base it applies. That is, speakers insist that the

preferred interpretation of m-reduplicated entries is negative.

With verbs, for example, the pejorative perspective entails that the action denoted by the

m-reduplicated entry has not in some way reached its full potential but rather was performed

superficially.

(20) Michelle Obama and Barack Obama are talking about Malia Obama’s exam for tomorrow. Michelle Obama says this to Barack Obama: Malia-∅ kitab-ı oku-du-∅ m-oku-du-∅ ama yarın… Malia-NOM book-ACC read-PST-3SG m-read-PST-3SG but tomorrow sınav-da zorlan-acak-∅. exam-LOC sweat-FUT-3SG

Lit.: ‘Malia read m-read the book but she will have hard times at the exam tomorrow.’

‘Malia skimmed through the book but she will have hard times at the exam tomorrow.’

Thus, Michelle Obama knows that her daughter studied a little bit but not enough, and she finds

  12  

a way to convey her perspective through the use of m-reduplication.

With adjectives, pejorativity entails some sort of flaw in the property denoted by the base

of the reduplicated entry. The speakers insist that the flaw is minor. Thus in (21), the flaw means

that there is some imperfection, based on the perspective taken by the speaker. The beauty

standards of female body have changed (at least in the public media) since the time Mona Lisa

was painted. A speaker sensitive to current media preferences for emaciated females could utter

(21) and still mean that Mona Lisa is pretty.

(21) Mona Lisa-∅ güzel m-üzel ama çok kilolu-∅. Mona Lisa-NOM pretty m-pretty but very overweight-NOM ‘Mona Lisa is pretty and the like but she is so much overweight.’ 3.4. Generalizations: the highlights

In this section, we briefly recap the observed patterns of m-reduplication.

Phonologically, the shape of the base restricts m-reduplication: if the base is inherently

m-initial, m-reduplication cannot apply.

Syntactically, m-reduplication is neutral with respect to lexical categories. As we have

seen in section 3.1, it freely applies across nominal, verbal, and adjectival domains. Moreover, it

applies to phrase level functors like pronouns, number, or quantifiers. However, m-reduplication

is not found beyond the phrase level: it is ungrammatical to m-reduplicate conjunctions or

complementizers, i.e. clause level functors.

Lexical restrictions apply to m-reduplicated utterances. Specifically, the lexical content of

the derived utterance must fit the superordinate category of the base. For example, a reduplicated

proper name of a person delimits m-reduplicated interpretations to humans. There is another

property that stands out in relation to lexical categories: default pejorative interpretation of m-

reduplicated entries.

Context loosens the interpretation of m-reduplicated utterances when the lexical base is

general enough. For example, an entry like ‘drink’ is inherently general in the sense that any

drinkable liquid could fit the overarching concept. Here, more flexibility is allowed with respect

to reduplicated utterance: ‘drink m-drink’ ranges over a variety of drinks, even including a side

of food or liquids that would not normally be considered drinks (e.g. blood).

  13  

Lastly, the use of m-reduplication is sensitive to register. While high frequency of m-

reduplicated utterances can be observed in colloquial contexts, we are yet to find instances of m-

reduplication in formal contexts.

4. Theoretical assumptions. Proposal

We have demonstrated the robustness and versatility of m-reduplication in Turkish. The

question that remains to be answered is: what is m-reduplication? In this section, we argue that

m-reduplication is a type of lexical number, namely, similative plural (4.2). Before we outline

the details of our proposal, we first spell out our theoretical assumptions (4.1).

4.1. Theoretical assumptions

It has been observed that expressions of number vary in form, meaning, and function

across and within languages (Corbett 2000, among many others).

Plurality is a kind of number, and recent studies also show that expressions of plurality

are not homogenous, (Ajiboye 2005, Butler 2012, Wiltschko 2008, among others). For the

purposes of this study, we assume that plurality may be functional or lexical, and that each

subtype would manifest specific properties. In particular, we assume that morphosyntactic

properties of lexical versus functional plurals can be reduced to head versus modifier distinction

identified and contrasted by Wiltschko (2008:689) in English (Germanic) and Halkomelem

(Salish), as summarized in Table 2 below.

Table 2. Functional versus lexical number English

functional number Halkomelem lexical number

Obligatory Yes no can trigger agreement Yes no absence associated with meaning

Yes no

can be selected for Yes no allows for form meaning mismatches

Yes no

Moreover, following Wiltschko (2008) and Ajiboye (2005), we assume that the syntactic

adjunction level of a modifier may also vary. I.e. it may adjoin as low as roots or higher, at

phrase level (the exact adjunction site varies with the specific language).

  14  

We further assume variation in lexical plurality. In particular, we focus on associative

plural (Moravcsik 2003, Daniel & Moravcsik 2013, among others). Associative plural refers to

human entities, meaning ‘X and X’s associated persons’, such as in Hungarian, illustrated below:

(22) Péter-ék Peter-APL ‘Peter and his family or friends or associates’ Moravcsik 2003:469 This type of associative plural has been discussed in the literature quite extensively (Daniel &

Moravcsik 2013, Kaneko 2013, Nakanashi & Ritter 2008, among others). This is not the case

with similative plural, which is a subtype of associative. Essentially, it is the same as associative,

except it may refer refers to entities that are non-sentient and unfamiliar4. Besides our Turkish

data, we have only seen one example of similative plural cited in the literature, from Telugu

(Dravidian), alas, without a gloss: puligili ‘tigers and such’ (cited in Daniel & Moravcsik,

2013)5.

In addition to our assumptions about the syntactic properties and lexical subtypes of

plurality, we also take a particular stance with respect to its distribution across categories.

Namely, we posit that plurality is not restricted to nominal domain alone, but rather may apply

across categories (Armoskaite 2010, van Geenhoven 2004, to name a few).

Scalarity is another theoretical tool we need to employ for the analysis of m-reduplicated

expressions. We assume that some linguistic expressions inherently introduce scales in their

interpretations, and that the strength of the expression with respect to a particular scale may vary,

e.g. on a scale of a timeline, the adverb always is stronger than the adverb often (Crystal

2008:423, among others).

4.2. Proposal: m-reduplication is a category-neutral lexical similative

We propose that Turkish m-reduplication is best accounted for as an instance of category-

neutral, similative plural. We show how its morphosyntactic structure is put together, and then

we explain how it is an example of a lexical rather than functional plural.

                                                                                                               4 We are grateful for an anonymous reviewer for pointing out that proper name pluralization may be an instance of associative in Turkish. Consider how ‘Ayşeler mayşeler’ differ in meaning from ‘Ayşeler’: a. Ayşeler mayşeler bize geldi. (Ayşe-PL m-Ayşe-PL us came) entails that when Ayşe came to us, she brought bunch of other people that we may or may not know. b. Ayşeler bize geldi. (Ayşe-PL us came) entails that Ayşe and probably Ayşe’s relatives that we are acquainted with came. Only the m-reduplicated example includes people beyond the intimate circle of Ayşe. 5 At this time, we can only hypothesize that Abkhaz (Bruening 1997) and Persian (Ghaniabadi et al 2006) may pattern similarly to Telugu.

  15  

Morphosyntactic structure. We have already observed in Section 1 that m-reduplication is a

structurally complex process. It happens in two steps: (i) reduplication (ii) m-prefixation. We

argue that each step is grammatically meaningful. We posit that the reduplication part encodes

plurality whose meaning is shaped by the lexical category of the base, while m-prefixation

encodes similativity6.

Our claim about reduplication encoding plurality is in line with some observations voiced

in the literature. For example, Ketrez (2012:260) briefly remarks: "Reduplication can also be

used as a measure of plurality or measurement". She then provides some examples that support

this view:

(23) a. bardak-∅ b. bardak-∅ bardak-∅ glass-NOM glass-NOM glass-NOM ‘glass’ ‘multiple numbers of glasses’

c. akın d. akın akın rush rush-NOM rush-NOM ‘sudden rush’ ‘rushing in crowds’

However, Ketrez (2012) gives just these few examples, but not within sentences. We tested

some more examples in our fieldwork, finding that reduplication of bare noun forms did indeed

induce collective meaning, as samples in (24)-(25) shows.

(24) a. damla-∅ b. damla-∅ damla-∅ drop-NOM drop-NOM drop-NOM

‘a drop’ ‘many drops, (generally refers to teardrops)’

c. Damla damla ağlı-yor-du. drop drop cry-PROG-PST-3SG

‘S/he was crying a lot (but crying calmly, not bursting into tears).’

(25) a. sıra-∅ b. sıra-∅ sıra-∅ row-NOM row-NOM row-NOM

‘a row’ ‘many rows’

c. Öğrenci-∅-ler tören-de sıra-∅ sıra-∅ yürü-yor-lar-dı-∅. student-NOM-PL ceremony-LOC row-NOM row-NOM walk-PROG-PL-PST-3SG ‘Students were walking in rows at the ceremony.’

                                                                                                               6 We thank Manfred Krifka (p.c.), for drawing this possibility to our attention.

  16  

Attributives - adjectives and adverbs- are discussed in more detail. It is noted that

productive partial reduplication is used as a means to denote ‘completely x’ or ‘x in large

measure’ (Ketrez 2012:260-261, Kornfilt 2000: 419; 435). We take it to mean that plurality,

when applied to non-count property, comes to mean ‘in large measure’. Thus, fully reduplicating

attributives allows us to introduce plurality of properties in the case of similative plural.

Note that when we say that reduplication encodes plurality, we mean that it is the source

of plurality for the entities, actions, or properties introduced by the reduplicated base, i.e. the

second element in reduplication. The first element, i.e. the base itself, remains ambiguous

between singular and plural interpretation in the case of nouns and is singular in the case of verbs

and attributives. Schematically, we can represent it as below:

(26) a. N m-N b. V m-V c. A m-A

[±singular] [plural] [singular] [plural] [singular] [plural]

As mentioned in Section 3.1 the nominal patterns reflect the general property of bare nouns in

Turkish, which can be denoted either plural or singular entities.

Thus, we have seen, in however few examples (23)- (25), that reduplication can be

attested on its own, without the m- prefix. We take it to mean that reduplication and m-

prefixation are two separate processes. We further argue that similative meaning arises with the

prefix m-. On one hand, there is nothing in the notion of plurality that would require it to also

necessarily be similative. Similativity is an additional feature. On the other hand, the

reduplicative examples did not carry similativity meaning, therefore, we conclude that

similativity is encoded by m-.

Now that we have discussed the morphological properties of m-reduplication, we turn to

its syntactic behavior. We argue that m-reduplication is an instance of lexical plural adapting

(Wiltschko 2008). The key syntactic property of a lexical plural is its status as modifier. In other

words, it is not obligatory, syntactically. This holds true of m-reduplication: it is always optional,

and we are yet to find an environment where m-reduplication is required.

Triggering agreement is one of the key properties of grammatical plural. In English,

plurality requires agreement or else the utterance is ungrammatical, as shown by the contrast in

(27).

  17  

(27) a. The cats like milk. b. *The cats likes milk.

Since cat in (27a) is in plural, the verb must agree in number. If the number agreement is lacking,

the utterance is ungrammatical, as in (27b). This is not the case in Turkish. Plural nouns can

surface either with singular or plural verbs, both grammatical, as the lack of contrast in

grammaticality shows:

(28) a. Kedi-∅-ler süt-ü sev-er-ler. cat-NOM-PL milk-ACC love- AOR -3PL ‘Cats love milk. b. Kedi-∅-ler süt-ü sev-er. cat-NOM-PL milk-ACC love- AOR.3SG ‘Cats loves milk.’    The same lack of contrast in grammaticality holds in m-reduplicated instances, as illustrated in (29): (29) a. Kedi-∅-ler m-edi-∅-ler süt-ü sev-er. cat-NOM-PL m-cat-NOM-PL milk-ACC love-AOR.3SG ‘Cats and the like loves milk. b. Kedi-∅-ler m-edi-ler süt-ü sev-er-ler. cat-NOM-PL m-cat-NOM-PL milk-ACC love-AOR-3PL ‘Cats and the like love milk.’

Thus, adding m-reduplication does not trigger agreement, and therefore we take it to mean that

m-reduplication patterns as lexical plural.

Another relevant property of a non-lexical plural is this: its absence should give rise to a

particular meaning. So, for example, the absence of plural –s in English is associated with

singular interpretation (book versus books). Thus, we know that English plural marking is

grammatical rather than lexical because its absence does result in a particular restricted meaning.

In contrast, the absence of m-reduplication does not cause a particular interpretation. The plural

similative/associative meaning is simply lost. This behavior of m-reduplication is akin to the

syntactic presence or absence of attributive modifiers in English. For example, an English

utterance a bright red apple contains two optional modifiers whose absence does not result in a

specific meaning if we say an apple.

  18  

M-reduplication is not selected for, which meets another criterion for lexical plural status.

For example, some English verbs require plural arguments (27ab) and some quantifiers require

plural nouns (30cd), i.e. select for plural entities:

(30) a. Students gathered for lunch. b. *Student gathered for lunch.

c. Three boys arrived. d. *Three boy arrived.

Given that in our Turkish data set we are faced with lexical number, one should not expect the

same selectional properties as manifested in English. In fact, one should not expect selectability

at all. And, to our knowledge, there is no linguistic expression that selects for m-reduplicated

entries in either nominal, verbal, or attributive domains.

In fact, one would expect to find precisely the opposite: restrictions of co-occurrence with

particular functional heads due to the specific lexical nature of m-reduplication where plurality is

necessarily tagged along with similativity. Indeed, such restrictions can be found. For example,

unlike simplex nominals (30a), m-reduplicated nominals cannot co-occur with numerals that

denote a precise number (30b).

(31) a. Ayşe’nin üç kedi-si var-∅. Ayşe’-GEN three cat-ACC have-3SG ‘Ayşe has three cats.’ b. *Ayşe’nin üç kedi-si m-edi-si var-∅. Ayşe’-GEN three cat-ACC m- cat-ACC have-3SG ‘Ayşe has three cats and the like.’ Akin to nouns, m-reduplicated attributives cannot co-occur with precision denoting functors, e.g.

with elements denoting the highest degree, such as en ‘most’. Thus, while simplex attributives

can inflect for the highest degree (33a,b), m-reduplicated attributes are rendered ungrammatical

(33c,d).

(33) a. Pırlanta-∅-lar pahalı-dır-∅. diamond-NOM-PL expensive-VBLZ-3SG ‘Diamonds are expensive.’

b. Pırlanta-∅-lar en pahalı-dır-∅. diamond-NOM-PL most expensive-VBLZ-3SG ‘Diamonds are the most expensive.’

  19  

c. Pırlanta-∅-lar pahalı-dır-∅ m-ahalı-dır-∅… diamond-NOM-PL expensive-VBLZ-3SG m-expensive-VBLZ-3SG ‘Diamonds are expensive and the like.’

d. * Pırlanta-∅-lar en pahalı-dır-∅ m-ahalı-dır-∅. diamond-NOM-PL most expensive-VBLZ-3SG m-expensive-VBLZ-3SG

Lastly, we argue that the default pejorative interpretation of m-reduplicated entries is

due to scalarity. We argue that scalarity is epiphenomenal in this case. It falls out of interaction

between plurality (reduplication), similativity (m-prefixation), and the lexical base. When

reduplication of the base occurs, then plurality is attained, and we have entities, events, or

properties. The derivation of meaning is not complete yet. The prefixation of m- results in

similativity, i.e., the reduplicated part is now interpreted not only as pluralized but also as being

similar to the entity, event, or property denoted by the base. A distinction exists between what

the two parts of the m-reduplication unit denote. The base of the m-reduplicated entry provides

the lexical content and thus serves as the conceptual foundation for the entities, events, or

properties denoted by the reduplicated, m-prefixed part. This distinction between the two parts of

the m-reduplicated phrase results in scalar tension. The base is the precise, ideal incarnation of

the lexical content while the similative bit is only similar to that ideal, but does not reach it

precisely because it is the inherent meaning of similativity. Moreover, plurality strengthens the

scalarity effect further in that there is more than one similar entity, event, or property that falls

short of the ideal denoted by the base. As a result of the tension between the base, reduplication

and similativity, a default pejorative meaning arises7.

5. Conclusions and further questions

In this study, we provided an in-depth description of the highly productive use of hitherto

little explored colloquial m-reduplication in Turkish. We presented a detailed inventory of its

uses across categories followed by a range of restrictions: phonological, semantic, pragmatic,

and stylistic. Given its semantic effects and lexical rather than functional properties, we proposed

that m-reduplication is an instance of lexical number. More specifically, we argued that m-

reduplication is a case of similative plural. Empirically, the study contributes to the growing

                                                                                                               7 While we are aware that the default preferred meaning is pejorative, we do not claim it to be final. The pejorativity is not fixed. One can manipulate the interpretation depending on a particular perspective of a particular speaker. For example, whether corpulent women are perceived as pretty or ugly may depend on individual tastes and cultural conventions, both of which are subject to change. We leave this issue for further research.

  20  

body of research on Turkish colloquial speech. Analytically, it offers tools to identify the finer

nuances of interaction between lexical number and similativity. Theoretically, it pushes the

boundaries of what we traditionally consider category number to be: namely, an obligatory,

usually inflectional functor that category specific. We show how data from a subset of Turkish

lexicon challenge the more traditional views, and pose new questions. For example, the puzzle to

be resolved is the category neutrality of m-reduplication. What is it that allows m-reduplication

to apply across categories? We hypothesize that the answer may lie in the semantics of m-

reduplication (see Gillon, Armoskaite & Kutlu 2014).

The study also raises a number of typological questions. For example, to the best of our

knowledge, only associative plural has been discussed in the literature in detail (Moravcsik 2003,

Daniel & Moravcsik 2013, among others), while similative is only mentioned (Daniel &

Moravcsik 2013) or its number-like behavior not considered (Müller 2004, Bruening 1997).

Moreover, we have noted that only nominals have been addressed with respect to associative

plural (Kaneko 2013, Nakanashi & Ritter 2008, among others). The robustness of the similative

pattern in Turkish begs to ask whether similar category neutral patterns have not been

overlooked in other languages, too. An exhaustive typology falls beyond the scope of this study.

In the future, we will explore similar reduplication patterns noted for Persian (cf. Ghaniabadi et

al. 2006) and Abkhaz (Bruening 1997).

References Ajiboye, Olaidipo 2005. Topics on Yorùbá nominal expressions. [Ph D dissertation, University

of British Columbia, Vancouver]. Armoskaite, Solveiga 2010. The destiny of roots in Blackfoot and Lithuanian. [Ph D dissertation,

University of British Columbia, Vancouver]. Bruening, Benjamin 1997. Abkhaz Mabkhaz: M-reduplication in Abkhaz, Weightless Syllables

and Base-Reduplicant Correspondence. In Bruening, Benjamin, Kang, Yoonjung, McGinnis, Martha (eds.) MIT Working Papers in Linguistics 30. Cambridge: Massachusetts. 291-329.

Butler, Lindsay 2012. The DP-adjoined plural in Yucatec Maya and the syntax of plural marking. Available online at http://www.hlp.rochester.edu/~lbutler/butler2012_dp_pl_yucatec.pdf

Corbett, Greville 2000. Number. Cambridge: Cambridge University Press. Crystal, David 2008. A dictionary of linguistics and phonetics. Oxford: Blackwell. Daniel, Michael & Moravcsik, Edith (eds.) 2014. The Associative Plural. In: Dryer, Matthew S.,

Haspelmath, Martin The World Atlas of Language Structures Online. Leipzig: Max

  21  

Planck Institute for Evolutionary Anthropology. (Available online at http://wals.info/chapter/36, Accessed on 2014-06-29.).

Hatipoğlu, Vecihe 1981. Türk dilinde ikileme. Ankara: Ankara Üniversitesi Basımevi. Ghaniabadi, Saeed & Ghomeshi, Jila & Sadat-Tehrani, Nima 2006. Reduplication in Persian. In the Proceedings of the 2006 annual conference of the Canadian Linguistic Association.

Available online at http://cla-acl.ca/?p=314 Gillon, Carrie & Armoskaite, Solveiga & Kutlu Deniz Aysegul. 2014. Plural domain widening

in Turkish. Ms, University of Rochester. Göksel, Aslı & Kerslake, Celia 2005. Turkish: a comprehensive grammar. London: Routledge. Görgülu, Emrah 2012. Semantics of Nouns and the Specification of Number in Turkish. [Ph D

dissertation, Simon Fraser University, Vancouver]. Nakanashi, Kimiko & Ritter, Elisabeth 2008. Plurality in languages without count-mass

distinction. Mass/Count Workshop handout, February 7-8. Müller, Hans-Georg 2004. Reduplikationen im Türkischen. Wiesbaden: Harrassowitz Verlag. Kaneko, Makoto 2013. Plural markers denoting salient and eventually intensional members in

Japanese. CNRS handout. Ketrez, Nihan 2012. A student grammar of Turkish. Cambridge: Cambridge University Press. Kornfilt, Jaklin 2010. Turkish. NY: Routledge. Moravcsik, Edith 2003. A semantic analysis of associative plurals. Studies in language 27.3,

469-503. Van Geenhoven, Veerle 2004. For adverbials, frequentative aspect and pluractionality. Natural

language semantics 12.2,135-190. Wiltschko, Martina 2008. The syntax of non-inflectional plural marking. NNLT 26.3, 639-694. Underhill, Robert (9th eds.) 2001. Turkish grammar. Cambridge: MIT Press.