CAPÍTULO 9. DISCUSIÓN Y CONCLUSIONES
9.5. P ALABRAS DE C IERRE
2.3.1. Principales investigaciones de Enseñanza Eficaz
After the critical discussion of the conceptual issues in Gutzmann (2015), the next destination is the data. This section will critically review the typology of UCIs proposed by Gutzmann (2013, 2015). Let’s recall the different types of UCIs, which we saw in the previous chapter:
4.1.2.2.1. Types of UCIs
f 2d r-s e.g.
isolated expletive − − ‘ouch’, ‘oops’
isolated mixed − + ethnic slurs, e.g. ‘Kraut’ functional expletive + − − ‘damn’
functional shunting + − + Japanese adverbial ‘yokumo’
functional mixed + + Japanese honorifics verbs, e.g. ‘irassharu’ (goHON) Table 4.1. Types of UCIs
f: functionality; whether the expression needs an argument
2d: 2-dimensionality; whether the expression has meaning in both dimensions (mixed) r-s: resource-sensitivity; whether the expression fully consumes its argument
As we can see, Gutzmann (2015) proposed the above typology based on two properties: dimensionality and functionality. However, using functionality as a criterion to distinguish between different subcategories of UCIs seems in itself problematic, given Gutzmann’s objectives. Even though he is clear that he defends ‘a conventionalist view of semantics’ (2015: 5), he states that one of the goals of his book is to ‘develop a compositional, multidimensional formal framework for hybrid semantics’ (2015: 9). Therefore if the mission is to provide an account of semantic composition, one would expect the property of interacting with other constituents (i.e. functionality) to be taken for granted, instead of allowing it to vary between plus and minus. After all, if some meanings always come saturated and cannot be arguments themselves either (due to
48 Crespo (2015) is here referring to Gutzmann’s (2016) ‘If expressivism is fun, go for it’, which
68 being ‘isolated’), one wonders what would count as good reasons for including them in the compositional process.49, 50 For Gutzmann, such a reason is conventionality.
As he notes, there are two possible criteria in order to delineate the domain of what one means by ‘semantics’: truth-conditionality and conventionality. This is in fact Kaplan’s (1999) point, and it has also been repeated by other philosophers, such as Kripke (2011) and Predelli (2013). However, it is not an accident that this point has been made by philosophers rather than formal semanticists; convention is a key notion in the discussion for a theory of meaning (ever since Plato’s Cratylus) but its inclusion in compositional semantics should not be automatically justified until it is clear that it participates in semantic composition. For example, apart from the obvious point of phonetic differences (such as accent), information about the speaker’s origins can be conveyed in a conventional manner by lexical differences across dialects (e.g. the use of ‘pop by’ and ‘mobile’ is conventionally perceived as British by US English speakers51). But it is not clear such information, albeit conventional, interacts with any other constituent in a compositionally relevant manner to yield meaning. On the contrary, it feels much more as a side-comment, or one that is added on top of everything else. Specifically for the examples just mentioned, if we represented the expressions’ dialectal profile in the semantic composition, we would be making the wrong predictions since whether something is perceived as dialectal or not usually depends on pragmatic considerations such as the speaker’s affiliation, the circumstances of discourse etc. (i.e. ‘flat’ and ‘mobile’ sound totally unmarked to ordinary British English speakers, although they could potentially perceive them as British-like in a communicative instance where another variety of English is spoken). Therefore it seems that conventionality cannot be a criterion for a definition of compositional semantics on its own, but needs to be constrained by something else to prevent our semantic representation from overflooding by too much information or becoming too pragmatic in nature. As noted above, one obvious constraint could be whether an expression takes an argument or can be used as an argument itself. If neither holds, it feels that there is no adequate justification for overcomplicating the semantics for representing such information there. Of course, this does not in any way mean that such information does not need to be represented; exactly because of its conventional character, expressions
49 To be clear, it is the fact that such expressions are both saturated and not argument material
either that points towards a treatment outside the compositional process. Just being saturated obviously won’t do, as the whole idea of Frege’s conjecture is to combine a saturated (such as a name or sentence) with an unsaturated component (Heim & Kratzer 1998). But we should also note that Gutzmann takes it that sometimes isolated UCIs can function as arguments, e.g. in constructions such as ‘Ouch, man!’ (2015: 92). I will discuss and try to refute this below.
50 As Potts (2005: 44) first noted when he defined the notion of isolated CIs, ‘when they [isolated
CIs] arise, the ‘side-comment’ nature of CIs is pragmatic: in a sense, the speaker is commenting on the circumstances of utterance, rather than expressing an aside about the at-issue comment’. However despite acknowledging their pragmatic nature, Potts (2005) opts for a purely semantic treatment of isolated CIs. For Amaral, Roberts & Smith (2008) this leads to inadequate formal treatment: ‘it seems that the isolated CI rule is a catch-all for phenomena which cannot be correctly accounted for by using CI application or COMMA’ (2008: 723-4). For this reason, the
present criticism against Gutzmann (2015) that isolated (marked as –functional) need not be represented in the grammar also extends to Potts (2005).
69 carrying such meaning deserve their own semantic profile52. Thus to clarify, the disagreement with Gutzmann (2015) is not whether conventionality is an adequate criterion for a theory of meaning, but whether it is also one for a theory of semantic composition. The immediate consequence here is that Gutzmann’s categories of isolated expletive and isolated mixed UCIs can no longer figure in a typology of UCIs, provided that the field is that of compositional semantics. Let’s see what implications these considerations have on the categories of isolated mixed and isolated expletive of Gutzmann’s UCI typology.
4.1.2.3. Banishing isolated UCIs