Shifting associativity:
A semantic and morphological analysis of
Tagalog Plural Pronoun Constructions
Jane Li
Johns Hopkins University
sli213@jhu.edu
Gérard Avelino
Rutgers University
gerard.avelino@rutgers.edu
AFLA 31. June 13, 2024
Associative plurals
When combined with some individual \(x\), yield "\(x\) and \(x\)'s family/friends/associates".
(Moravcsik 2003)
Turkish (Göskel & Kerslake 2005)
ablam -ler
sister.SG
-APL
"my sister and her family"
WALS 36: The Associative Plural (Daniel & Moravcsik 2013)
Theoretical
How is this dimension of meaning realized (in form) and composed (in meaning)?
How does Tagalog morphologically encode APL? Can we provide a morphological + compositional semantic analysis of it?
Empirical
Strategy 1: plural case marking
Tagalog expresses associative plurality with plural case marking, even when the DP is morphologically singular.
(1) Nakita ni John si Maria
see.PST
NOM.SG
Maria
"John saw Maria."
GEN.SG
John
(2) Nakita ni John sina Maria
see.PST
NOM.PL
Maria
sina
"John saw Maria and her associate(s)."
GEN.SG
John
Strategy 2: plural pronoun constructions
Another way Tagalog expresses associative meaning is through plural pronoun constructions (PPCs).
General form: plural pronoun [GEN NP]
(3) Nakita ni John kami
see.PST
1P
GEN.SG
John
"John saw us."
(4) Nakita ni John kami ni Maria
see.PST
1P
GEN.SG
Maria
John
GEN.SG
lit. "John saw Maria's us."
"John saw Maria, me (, and maybe someone else)."
Inferences that these constructions introduce
⇝ John saw Maria.
⇝ John saw Maria.
1
(2) Nakita ni John sina Maria
see.PST
NOM.PL
Maria
sina
GEN.SG
John
Plural case marking
m
⇝ John saw at least one of Maria's associates.
⇝ John saw at least one of Maria's associates.
⇝ I am one of Maria's associates.
(4) Nakita ni John kami ni Maria
see.PST
1P
GEN.SG
Maria
John
GEN.SG
kami ni Maria
Plural pronoun construction
m
Why Tagalog?
Previous analyses have focused on languages with strategies like Turkish (Nakanishi & Tomioka 2004, Vassileva 2005, Smith 2020, New 2021). Analyzing these alternative strategies in Tagalog helps us understand the morpho-syntactic nature of associative plurality.
Semantic work on PPCs have focused on deriving the numerical inferences (Russian: Vassileva & Larson 2005, Tlingit and Polish: Cable 2017; Icelandic: Sigurðsson & Wood 2020) but not the associative meaning (though see Yuan 2017).
Interaction between the two constructions are possible! Further pinpointing perhaps available?
Questions for today
Assumption: associative plurality is a morpheme (APL) and across the two strategies, share the same source within the DP.
- What type of morphology needs to happen such that we get the two constructions on the surface?
- What is the lexical semantics of APL? How does it compose with other elements (e.g., Maria)?
- Specifically, for PPCs: how do we incorporate the inference that the person presupposed by the pronoun (1st, 2nd, ...) is a part of the individuals inferred?
Synopsis / Outline
- Association is a projection within the DP. It has an uninterpretable plural feature. Case looks for the plural feature on Num, hence plural case marking.
- If a person feature is there, this is all morphologically realized as a plural pronoun. Semantically, this pronoun is type <\(e,\, e\)>, forcing the construction to have a complement ni Maria.
- Association is a projection within the DP. It has an uninterpretable plural feature. Case looks for the plural feature on Num, hence plural case marking.
- We will interleave the morpho-syntax analyses with the semantics. It works, but there are caveats.
- The lexical semantics nicely predicts a bunch of things, but there is much more to be done.
AssocP and NumP
- Based on evidence from Turkish, Japanese, and other languages, Cinque (2018) posits that there is a projection AssocP above NumP.
- Lewis (2021) follows up by positing that Assoc has an uninterpretable plural feature that forces head movement + adjunction from Num. This explains why so many languages have the same morphological marker for APL and additive plural.
[iAssoc]
[uPL: _ ]
[iPL]
AssocP
KP
NumP
...
...
...
NumP
NP
K
Assoc
Num
AssocP
KP
NumP
...
...
...
NumP
NP
K
Assoc
Num
[iAssoc]
[uPL: PL]
\(t\)
Assoc
Num
[iPL]
AssocP and NumP
- Based on evidence from Turkish, Japanese, and other languages, Cinque (2018) posits that there is a projection AssocP above NumP.
- Lewis (2021) follows up by positing that Assoc has an uninterpretable plural feature that forces head movement + adjunction from Num. This explains why so many languages have the same morphological marker for APL and additive plural.
AssocP
KP
NumP
...
...
...
NumP
NP
K
Assoc
Num
[iAssoc]
[uPL: _ ]
[iPL]
KP
NumP
...
...
...
NumP
NP
K
Assoc
Num
[iAssoc]
[uPL: PL]
\(t\)
Assoc
Num
[iPL]
AssocP
Turkish
(Göskel & Kerslake 2005)
ablam -ler
sister
-APL
"my sister and her family"
ev -ler
house
-PL
"houses"
Deriving strategy 1: plural case marking
- Instead of overtly realizing the plural morpheme like other languages, Case does its thing and checks for number and proper/common noun on N.
sina
- Realizes phonologically the appropriate case here: sina.
AssocP
KP
NumP
...
...
...
NumP
NP
K
Assoc
Num
[iAssoc]
[uPL: PL]
\(t\)
Assoc
Num
[iPL]
NOM
Proper noun
Outstanding questions:
- Semantics?
- Unattested: NP is common noun.
Semantics of strategy 1: plural case marking
AssocP
NumP
...
...
NumP
NP
Assoc
Num
[iAssoc]
[uPL: PL]
\(t\)
Assoc
Num
[iPL]
Proper noun
\([\![\text{PL}_{+\textrm A}]\!] = \lambda x. \sigma y[\ast\text{Assoc}(x)(y)] \oplus x\)
\([\![\text{PL}_{+\textrm A}]\!]([\![ \text{Maria}]\!])= \sigma y[\ast\text{Assoc}(m)(y)] \oplus m\)
\([\![\text{PL}_{+\textrm A}]\!]\) is looking for an individual (type \(e\)). If NP is a common noun, its type should be \(\langle e,\,t\rangle\) at this point. Type clash rules out the possibility of having a common noun as the NP.
Deriving strategy 2: PPCs
Person features within the DP force it to spell out as the plural pronoun.
AssocP
KP
NumP
...
...
...
NumP
NP
K
Assoc
Num
[iAssoc]
[uPL: _ ]
[iPL]
NOM
Deriving strategy 2: PPCs
Person features within the DP force it to spell out as the plural pronoun.
AssocP
KP
NumP
...
...
...
NumP
NP
K
Assoc
Num
[iAssoc]
[uPL: _ ]
[iPL]
Pers
[i1]
PersP
NOM
Outstanding questions:
- Why do we need ni Maria?
- Semantics?
kami
Semantics of strategy 2: PPCs
AssocP
KP
NumP
...
...
NumP
K
Assoc
Num
[iAssoc]
[uPL: PL]
\(t\)
Assoc
Num
[iPL]
...
NP
Pers
[i1]
PersP
NOM
\([\![\text{PL}_{+\textrm A}]\!] = \lambda x \lambda y. \sigma z[\ast\text{Assoc}(x)(y)\wedge x\leq z] \oplus y\)
\([\![\text{PL}_{+\textrm A}]\!]^g([\![1]\!]^g) = \lambda y. \sigma z[\ast\text{Assoc}(g(1))(y)\wedge g(1)\leq z] \oplus y\)
Caveats:
- This \([\![\text{PL}_{+\textrm A}]\!]\) is different than the previous one!
- Need some sort of mechanism to interpret Person with the bundle first.
Semantics of strategy 2: PPCs
AssocP
KP
NumP
...
...
NumP
K
Assoc
Num
[iAssoc]
[uPL: PL]
\(t\)
Assoc
Num
[iPL]
...
NP
Pers
[i1]
PersP
NOM
\([\![\text{PL}_{+\textrm A}]\!] = \lambda x \lambda y. \sigma z[\ast\text{Assoc}(x)(y)\wedge x\leq z] \oplus y\)
\([\![\text{PL}_{+\textrm A}]\!]^g([\![1]\!]^g) = \lambda y. \sigma z[\ast\text{Assoc}(g(1))(y)\wedge g(1)\leq z] \oplus y\)
Another semantic argument:
\([\![\text{PL}_{+\textrm A}]\!]\) is looking for an individual (type \(e\)). If the pronoun is left alone and we assume that at KP types should be \(e\) (or something in the type-shifting triangle), there will be a clash at KP. Therefore, there must be an \(e\) argument.
Taking stock
- Our analysis falls out from existing theories of associative plurality with other strategies (Cinque 2018, Lewis 2021).
- While our semantics works for these two constructions, their semantics is slightly different, and that is not desirable. However...
- This is done with a lot of help from assuming that things will morphologically fuse.
Formal connections?
\(\lambda P. \sigma y[\ast P(y)]\)
\(\lambda x \lambda P. \sigma z[\ast P(x)\wedge x\leq z]\)
| Associative | Non-associative | |
|---|---|---|
| Reduced |
|
|
| Non-reduced |
|
Definite plural (Kurafuji 1999):
the teachers
Adnominal pronoun constructions:
we linguists
\(\lambda x. \sigma y[\ast\text{Assoc}(x)(y)] \oplus x\)
\(\lambda x \lambda y. \sigma z[\ast\text{Assoc}(x)(y)\\\wedge x\leq z] \oplus y\)
Case marking:
sina Maria
PPCs:
kami ni Maria
Beyond the core data
- The analysis directly extends to 1st person inclusive and 2nd person plural pronouns.
- But there are some problems with 3rd person (ask us about it!).
⇝ You are one of Maria's associates.
2
m
⇝ John saw Maria.
⇝ John saw at least one of Maria's associates.
(5) Nakita ni John kayo ni Maria
see.PST
2P
GEN.SG
Maria
John
GEN.SG
Beyond the core data
- The analysis directly extends to 1st person inclusive and 2nd person plural pronouns.
- But there are some problems with 3rd person (ask us about it!).
- It also extends to cases where the overt referent (e.g., Maria) is now multiple named people (e.g., Maria at Carlos "m and c").
- PPCs can also take common nouns as overt referents, but only when they are strongly definite (kami ng guro "me+teacher(+others)").
- Stacking constructions...
An interaction between the two constructions
⇝ John saw Maria.
⇝ John saw at least two of my associates.
⇝ Maria is one of my associates.
⇝ John saw Maria.
⇝ John saw at least one of Maria's associates.
⇝ I am one of Maria's associates.
m
1
(4) Nakita ni John kami ni Maria
see.PST
1P
GEN.SG
Maria
John
GEN.SG
(6) Nakita ni John kami nina Maria
see.PST
1P
GEN.SG
John
GEN.PL
Maria
1
m
An interaction between the two constructions
(4) Nakita ni John kami ni Maria
see.PST
1P
GEN.SG
Maria
John
GEN.SG
(6) Nakita ni John kami nina Maria
see.PST
1P
GEN.SG
John
GEN.PL
Maria
\([\![\text{PL}_{+\textrm A}]\!]^g([\![1]\!]^g)\)
⇝ John saw Maria.
⇝ John saw at least two of my associates.
⇝ Maria is one of my associates.
⇝ John saw Maria.
⇝ John saw at least one of Maria's associates.
⇝ I am one of Maria's associates.
m
1
1
m
\(([\![\text{Maria}]\!]^g)\)
\([\![\text{PL}_{+\textrm A}]\!]^g([\![\text{Maria}]\!]^g)\)
\(([\![1]\!]^g)\)
Future work
Conclusions
- What is the semantic contribution of the genitive marker itself?
- Genitives as modifiers (Partee & Borschev 1998, 2001, 2003), a way to assert an asymmetric relation in general?
- Tension with PPCs of other languages, which don't use the genitive?
- Scoping and binding behaviour.
- We provide an syntactic/morphological analysis that falls out from existing theories of associative plurality, with some additional assumptions to explain morphemes are spelled out in Tagalog.
- The semantics built on top of these morphological assumptions also works for the most part, and shows interesting parallels with other DP constructions.
Thank you! 🌻
Our heartfelt thanks to the JHU Semantics lab and the Rutgers SURGE reading group for suggestions.
afla31-li-avelino
By janeli