by David J. Lobina
It is often stated that natural language is both ambiguous and inexplicit to be the medium of thought – to be the medium in which we think, as opposed to the conceptual language of thought I have gone on about for some time here at 3 Quarks Daily. But what do we mean when we say that language is ambiguous and inexplicit?
In the case of ambiguity, there are two relevant cases: lexical ambiguity, where a word may mean different things in different contexts (think of the word bank and all the things it can refer to); and structural ambiguity, as in the sentence below, which can receive different interpretations depending on what part of the sentence the phrase in Paris actually modifies.
The author of Stephen Hero decided to write Finnegans Wake in Paris.
That is, did the decision to write Finnegans Wake take place while the noted author was in Paris, or did the author of Stephen Hero decide that Paris would be a suitable place in which to carry out the foreseen writing? Or to put it in more linguistic terms: does the prepositional phrase in Paris modify the overall phrase decided to write Finnegans Wake or just the internal and thereby shorter phrase to write Finnegans Wake? Given the differing interpretations, such a sentence couldn’t possibly constitute an object of thought, for there is no such thing as an ambiguous thought. How could a person even have an ambiguous thought?
Indeed, whoever utters such a sentence is in no doubt as to what they mean by it, and that’s not only because an unambiguous thought has been entertained. The utterer has, in addition, internally constructed a syntactic object of the right hierarchical structure, a construction that necessarily precedes producing the string of words, one by one, that the hearer or reader actually experience in a communicative interaction.
The structural ambiguity only arises for the hearer, and this is so only on account of the manner in which (verbal) communication is conducted – in a segment-by-segment basis. But there are two syntactic objects associated to the string above; one’s linguistic system generates two different structures for the same set of words. Thus, what is linguistically represented in the mind of a speaker in the case of the sentence above are two syntactic objects respecting the hierarchical relations laid out, simplifying significantly, below, where the bracketing and underlining are supposed to mark the right dependencies.
The author of Stephen Hero [decided [to write Finnegans Wake] [in Paris]].
The author of Stephen Hero [decided [to write Finnegans Wake in Paris]].
If this is so, then the case of structural ambiguity in language and thought is a bit more nuanced, given that the language faculty generates unambiguous structures, and as such language is, in this case at least, perfectly capable of appropriately differentiating the contents, or thoughts, of two different conceptual representations. It just so happens that the linguistics representations become ambiguous in the process of converting hierarchical structures into an appropriate signal for communication – linear strings of words – but this an altogether different matter.
The issue of whether language is explicit enough to account for the thoughts we do have on a regular basis possibly carries more weight. This is the claim that very often language is rather unspecific regarding what content it expresses; or put differently, that linguistic phrases say less than what a speaker is mentally representing in their mind at the point of utterance.
Descriptions such as the author of Stephen Hero, to keep with the example at hand, can indeed be very vague regarding who they refer to, and as such these expressions can be rightfully considered to be not explicit enough, as the underlying representation would be composed of (or being connected to) a much richer conceptual representation. The point is not that one couldn’t entertain a concept such as AUTHOR OF STEPHEN HERO in itself and on its own (concepts in capital letters, as philosophers and psychologists write them).
That is to be granted, of course; the point is that we must also allow for a situation, plausibly a much more common one, in which one knows the name of the author full well but chooses to refer to them by using a definite description. In such circumstances, we would want to say that a speaker is entertaining many different concepts in addition to AUTHOR OF STEPHEN HERO – not least, in this case, the concept JAMES JOYCE – but chooses to use the phrase the author of Stephen Hero for reasons other than what they are simply mentally representing. After all, AUTHOR OF STEPHEN HERO is but one among the many other pieces of information a given person may know about James Joyce the person (among others, that he hailed from the island of saints and sages).
Using the phrase the author of Stephen Hero must obey other concerns, possibly to do with communication – to sound clever, to be purposely cryptic, or else – but the thought one is entertaining at the time of uttering that description is necessarily something else, and something clearly more expanding. That is, one would be entertaining the concept JAMES JOYCE and whatever surrounds it, choosing one of these orbiting pieces of information to construct the eventually externalised linguistic phrase – in the case at hand, the relevant datum would be the not so well-known fact that James Joyce wrote an autobiographical novel by the name of Stephen Hero. In this sense, it is not only that a linguistic expression rarely transparently identifies what concept it express exactly; language can furthermore be quite misleading as to what conceptual representations is related to.
Linguistic structure can only go some far towards establishing the meaning of a sentence, and extralinguistic factors need to be taken into consideration; among other things, context, the perceived intention of a speaker, and conceptual information in general (by the latter I mean the overall general and encyclopaedic knowledge a speaker possesses, and how this information interacts with concepts and the words of a language).
A dramatic case of what I have in mind here has to do with so-called deictics, words than receive a direct referential interpretation – proper names like James Joyce and demonstratives such as this, for which one must ascertain what they refer to precisely in order to work out their meaning. The phenomenon is in fact rather widespread and concerns the deictic uses of language, the most obvious cases involving pronouns such as he, we and the like (including the just-mentioned demonstrative pronouns), but also present in, inter alia, the language of time (tomorrow, later, etc.) and space (here, there, etc.).
The idea is that deictic words and phrases can only be fully interpreted when supplemented with contextual information, where the context may involve general knowledge, background knowledge, discourse knowledge, or the mutual knowledge that interlocutors may share. As in the case of structural ambiguity, the task of working out what these elements refer to in a given context falls squarely on the hearer, and this would indicate that here too what the speaker is conceptually representing whilst using such words is something else.
The typical state of affairs hearers may find themselves in is most dramatically illustrated in the case of sentences such as he decided to write it there later, which contains a number of purely deictic words (he, later, etc), and where the speaker is being a bit too economical with the linguistic output. Uttering such a sentence may certainly be cryptic on occasion (and not very stylish to boot), but this would usually be done on the assumption that the hearer is in possession of the (ample) knowledge needed to deduce what person, piece of writing, place, and time the speaker is pointing to – pointing being the essence of deixis, in fact.
In doing so, the hearer is enriching the content of the sentence so received, internally generating a conceptual representation that is more inflated than what the sentence directly conveys, and perhaps one that would be close enough to what the speaker was mentally entertaining. The general situation in which speakers and hearers find themselves is worth emphasising. The context may well help the hearer in fixing the references of each deictic element in a sentence – and semanticists might be satisfied with doing that in their theories too – but the speaker doesn’t require any contextual cues whatsoever to fix their thoughts, as these would be unequivocal and quite often in no need to be communicated at all.
The propositional enrichment hearers conduct during verbal communication brings me to two interesting issues. Firstly, it doesn’t appear to be the case that thought makes use of analogue concepts of deictic words and phrases. That is, the (re-)constructed conceptual representations a hearer puts together would not include concepts such as HE, THERE, LATER, etc, as there probably aren’t such concepts to begin with.
Instead, each and every one of these deictic terms would receive a rather different type of representation in thought – a more determinate representation, that is. In fact, it is not at all difficult to take any expression containing deictic terms and turn it into a more apt and explicit representation for the purposes of analysing the underling thought, and I would say that this is simply what typically happens in mental life. This shouldn’t be surprising, in fact, for the act of pointing is something one does (with words or else) – that is, it is a type of (in this case, linguistic) behaviour – and certainly not something that thought itself does.
Granted, one’s attention can be drawn to very many things – to that troubled artist over there, for instance – but the thought thus had would not be quite like that; once someone has directed their attention to that troubled artist over there, the concept being entertained would simply be TROUBLED ARTIST (or what have you) – and the same applies to definite articles and the like.
What I am suggesting, then, is that the thoughts underlying linguistic expressions must be propositional in nature (or at least close enough), which brings us back to all those entries on the language of thought, and as such they would not contain deictic terms, a constraint that would in addition rule out many other word types from having an analogue in thought.
That much must be true, I think, but we also have to consider the non-referential uses of deictic words, as in James Joyce once met THIS troubled artist, and he told him that…, where the demonstrative this is not referring to a specific individual; rather, it is being used as an indefinite article (a/an). Would non-referential uses of deictic words and determiners such as indefinite articles have a place in the LoT, then? I think this possibility is also rather unlikely, and for much the same reasons I have banished the referential kind of determiners from conceptual representations – in thought the primitive elements would be of a different kind. In the case of an indefinite article, and this applies to indefinite uses of demonstratives too, I would say that the corresponding thought representation would contain the concept ONE in its place, and I suspect something along these lines is the case for many other demonstratives and determiners.
What I am outlining, effectively, is a process of propositional enrichment, where deictic terms are converted into fuller conceptual representations during the language comprehension process. This brings us to a much larger issues, and a more complex one, which is the formation of inferences during the comprehension of language, and the domain of pragmatics properly. This follows from the very basic fact that Hearers extract a lot more information from the signal they receive than is actually encoded in it (the latter would be the domain of semantics), and they do so in a way that may go beyond constructing full propositions for the sentences a speaker utters, the issue I have focused on here.
Namely, the conceptual representations hearers construct for the input sentences must capture the information structure so crucial in linguistic communication, an aspect of thought that is codified in a particular type of representation. I’m referring to such properties as a sentence’s focus and presupposition(s), or in a more recent reformulation, a sentence’s topic and comment – that is, what the sentence is about and what is predicated about it, respectively. As such, what hearers effectively extract from the signal are topic-comment structures, a specific type of propositional structure that must be entertained in the language of thought I constantly bang on about – perhaps as a subtype of the canonical predicate-argument structure of propositions that I prefer – but this is a topic for another day.[i]
[i] I do not forget that I promised to come back to the contentious debate around the uses and abuses of the term fascism to describe some current events, especially in the US, and I will indeed come back to this next month.