LF impoverishment in DM
and polite addressee pronouns
SUMMARY: Couched in the framework of DM (Distributed Morphology), this paper derives the
typology of polite addressee pronouns with the help of ‘LF impoverishment rules’. This paper argues that LF impoverishment deserves more attention than the received view, thus updating DM with new typological data.
PUZZLE: In Mandarin, the personal pronoun nín means ‘you’ and it’s used to address a single person politely and it only has this function. I call this type of pronouns ‘dedicated polite addressee pronouns’. Conversely, in French, the pronoun vous also means ‘you’, but it can not only be used to address a single person politely but also a plurality of people in all possible scenarios. This is a case of syncretism and the derivational process of such a pronoun can be roughly modelled by proposing that there is an LF impoverishment process that deletes a plural number (PL) feature:
(1) French: (source form) 2nd pl(target form) 2nd polite sg
a. Syntactic input: {D HON PL...}
(HON for ‘honorificity’)
b. PL→/[__HON] (Impoverishment rule at LF) c. {D PL...}/vu/ (Vocabulary Insertion rule at PF) For convenience, I describe the uninterpretable PL in (1b) as ‘marking politeness’ because this feature is associated with HON feature which gives rise to politeness. In fact, based on the current typology (ca. 70 languages or so) that I gathered from WALS and my own data pool, a good range of features ‘mark’ politeness and I hypothesize that these features are also targets of deletion triggered by HON, as (1b) shows. Apart from the French pattern, where second person plural pronouns are used as second person singular polite pronouns, (2-6) shows that, together with (1) and ‘dedicated polite addressee pronouns’ as in Mandarin, polite addressee pronouns have at least 8 patterns, in which other features like dual number (DU), paucal number (PAUCAL), feminine (FEM) and reflexive (REFL) also ‘mark’ politeness.
(2) Muna (Van den Berg 1989): (source form) 1st incl(target form) 2nd sg polite
a. Syntactic input: {D HON DU...}
b. DU→/[__HON] (Impoverishment rule at LF) c. {D DU ...}/intaidi/ (Vocabulary Insertion rule at PF)
(3) Modern Standard German: (source form) 3rd pl(target form) 2nd sg polite a. Syntactic input: {D HON PL...}
b. PL→/[__HON] c. {D PL... }/zi:/
(4) Italian (source from) 3rd sg fem(target form) 2nd sg polite
a. Syntactic input: {D HON FEM...}
b. FEM→ / [__ HON]
c. {D FEM...}/le/ (Vocabulary Insertion rule at PF)
(5) Fijian: (source form) 2nd plural/paucal/dual(target form) 2nd sg polite
a. Syntactic input: {D HON PL/PAUCAL/DU...}
b. PL/PAUCAL/DU1 → / [__ HON] (Impoverishment rule at LF) c. {D DU...}/(o)mudrau~(o)drau/
or {D PAUCAL...}/(o)mudou~(o)dou/
or {D PL...}/(o)munnu~(o)nuu/ (Vocabulary Insertion rule at PF)
1 The LF rules for DU and PAUCAL impoverishment are an oversimplification here. DU and PAUCAL are assumed to be more complex number features than PL and could presumably be made of more than one sub- features. For example, in Ackema and Neeleman (2018), DU may consist of AUG (for ‘Augmented’) and MIN (for ‘minimal’). In Harley and Ritter (2002), PAUCAL is made up of AUG, MIN and GROUP.
I assume that one HON feature triggers one impoverishment rule of one syntactic feature at a time. For pronouns of complex feature values like DU and PAUCAL, HON features could be stacked: HON-HON, or HON-HON-HON, which could trigger the deletion of two or three syntactic features. In fact, the stacking of HON features has its empirical basis, which is omitted here due to limited space.
(Impoverishment rule at LF) (Vocabulary Insertion rule at PF)
(Impoverishment rule at LF)
(6) Hungarian: (source form) anaphoric reflexive pronoun(target form) polite 2nd sg
a. Syntactic input: {D HON REFL (PL)...}
b. REFL → /[__ HON] (Impoverishment rule at LF) d.{D REFL PL}Önök or {D REFL}Ön (Vocabulary Insertion rule at PF) It is not attested in my typology that case features can ‘mark’ politeness in addressee pronouns. A priori, one would think that using feature(s) which has no interpretation to begin with to mark politeness is the optimal strategy. For example, case features. If one adopts this route, then there is no need to have special rules post-syntactically to delete politeness marking features at LF, and any potential disparities in form and interpretation are avoided. The fact that a variation of other kinds of features like traditional phi features and REFL can be used makes it all the more interesting to ask why case features are not used in marking politeness in polite addressee pronouns. Phi features, after all, is but a cover term for person, number, and gender features.
PROPOSAL: I hypothesize that the key to the puzzle has to do with the internal distribution of pronominal features and the locality condition on the licensing of HON feature, constraining how potential politeness marking features would interact with HON in the syntax within a pronoun. Specifically, I argue that case features cannot interact with HON in the syntax due to the fact that access of HON to case features are ‘blocked’ by the DP/NmaxP2 phase boundary. HON can interact with phi features and other DP internal features like REFL because these features are within the DP/NmaxP phase. I relativise this ‘interaction’ to a licensing condition (7).
(7) The Threshold Requirement: HON triggers deletion of features that HON is dependent on, after being licensed and rendered active via Agree-Link3 by features that HON selects in the syntax.
LF Impoverishment: ‘Impoverishment’ normally refers to post-syntactic deletion rules at PF before phonological interpretation. ‘LF impoverishment’ is much less well known than PF impoverishment. However, LF impoverishment, namely deleting features post-syntactically before semantic interpretation, has already started to emerge since at least Grimshaw (1997), her idea is that do has semantic content, but it’s very minimal. A do-support do, however, does not have semantic content at all, which equals suggesting that there is LF impoverishment of the content of do. Heim (2008) discusses a type of phenomenon that could be potentially analysed as a case of LF impoverishment: phi features on bound pronouns. In a sentence like Only I did my work (Heim, 2008), the 1st person feature on bound pronoun my is not interpreted when this sentence means ‘x is such that x is the only person that did x’s homework and x refers to the speaker’. The non-interpretation of 1st person feature on the bound pronoun my could be potentially seen as undergoing deletion of 1st person (or [+author, +participant]) feature. Nevins (2008) explicitly identifies LF impoverishment as a parallel deletion process compared to the PF impoverishment in a Y-model grammar and termed LF impoverishment ‘deprivation’. He gave an example from the language Santali, in which the first person inclusive is used as second person singular with a threatening pragmatic effect towards the addressee. The analysis implied in Nevins (2008) is that an LF impoverishment rule changes the features for 1st person and non-singular number into 2nd person and singular.
Further PREDICTIONS: This paper not only accounts for attested polite addressee pronouns, but also predicts impossible polite addressee pronouns. Apart from hypothetical case-marked polite addressee pronouns explained in this abstract, this paper further predicts that 1st person singular pronouns, 1st person exclusive pronouns, masculine personal pronouns (in a 2-way gender system) and 2nd person singular pronouns cannot be recruited in any language as polite addressee pronouns, which is substantiated by the current typology.
Future EXPERIMENT(s): I assume that it is a surprising bias that the majority of languages that have a polite distinction in addressee pronouns are like that of French (1). Inspired by Jennifer Culbersten and colleagues’ work (eg. Culbertson and Kirby, 2016), this paper proposes that it has to do with a learning bias which is strengthened by cultural factors: simplicity.
Selected References: Ackema & Neeleman 2018. Arregi & Nevins 2012. Chung 1998. Culbertson and Kirby 2016. Grimshaw 1997. Heim 2008. Nevins 2008.
2 NmaxP is a label for the complement of D head.
3 Chung (1998) first proposed a two-step model of agreement. Arregi and Nevins (2012) call the first step ‘Agree-Link’ and the second step ‘Agree-Copy’.