Reported Speech and viewpoint hierarchy

Spronck & Nikitina’s (S&N) paper outlines a range of linguistic facts pertaining to the phenomenon widely known as Reported Speech (RS). S&N are right in observing that the selection of forms is typically very diverse, even if we consider one language, and becomes even more complex from a cross-linguistic perspective. S&N look at the phenomenon very broadly, while maintaining a consistent definition, and this in itself makes the article useful to any linguist interested in RS, or, more generally, in perspectivization. Also, a broad perspective is what linguistic typology needs – unifying rather than dissecting. And yet I would like to mention some dimensions of analysis which merit further discussion.

First, I am not sure that it is indeed the case that linguists typically agree on what does and does not count as reported speech, since descriptions offered range between quite narrow and very broad, especially if we include narrative and stylistic considerations. Many linguists might agree on a definition whereby one subjectivity (usually referred to as ‘the speaker’) communicates his or her representation of another subjectivity’s view of a situation (not necessarily expressed through speech, as S&N admit early on in the argument). However, syntactically speaking, instances of RS may be extremely varied, including examples of ‘depiction’ or ‘demonstration’ (Clark & Gerrig 1990; Clark 2016), which may not ‘feel like’ instances of RS, and yet are (rightly) considered by S&N to represent the same RS domain. The question is, then, why we should insist on referring to this broad spectrum of instances as a ‘syntactic domain’, given that the phenomena within the scope of the definition are not quite syntactically cohesive, while being cohesive semantically.

In the same vein, I would paraphrase what S&N say, and argue that reported speech constructions constitute a distinct class (we agree here), characterized by perspectival meaning. That is, I would delete the ‘proneness to’ part of the description by S&N, and just assume perspectivization to be the defining feature. However, I would also want to nuance the issue of viewpoints expressed. It is important to keep in mind that S&N see bi-perspectival meaning in the so-called ‘current speaker’ situation – whoever is reporting, is perceiving what is being reported and evaluating it too. I would add that in most cases there is also the viewpoint of the so-called ‘represented speaker’, and this viewpoint is embedded in the representation provided by the ‘current speaker’ (see Dancygier & Vandelanotte 2016 for more discussion of viewpoint configurations). In other words, the revised definition would recognize ‘a dual higher viewpoint on the lower viewpoint’ (elsewhere, I described such constructional phenomena as ‘stance-stacking’, Dancygier 2012b). Having looked at the examples that I have previously considered and then at the very informative overview of cases in the S&N paper, I am convinced that such a viewpoint hierarchy is the core of this domain – but if so, why make syntax primary, rather than describe an impressive variety of expressions as representative of a general perspectivization pattern?

Going further in that direction, I would very much welcome a more nuanced view of how the variety of RS forms constitutes a challenge to a construction grammar interpretation. S&N talk about ‘constructions’, but the question remains how such constructions are to be described. Moreover, the widely accepted definition of a construction as a relatively stable form-meaning pairing does not quite match the spectrum of cases included under the rubric of RS. My personal attempt to look at RS (in narratives written in English, so by definition in a more narrow context, cf. Dancygier 2012a) led me to argue that we could learn something by studying tense, or pronouns, or clausal structure, or sequence of clauses (if there are two), or textual embedding, and then looking at how combinations of these forms yield meaning, but that postulating a family of related constructions does not answer all the questions. I concluded, as a result, that the constructional approach is only distracting us from the real question – the nature of perspectivization in various RS forms. What S&N seem to be doing, in this context, is postulating a syntactic generalization on the basis of a semantic one.

To further explain the role of viewpoint configuration in the meaning of RS forms I want to briefly consider one specific parameter, the choice of the first person pronoun I, and show how it is interpreted in various narrative contexts. In general, first person fictional narratives rely on the dominant dual viewpoint which combines the roles of the character and the narrator. But further viewpoints can be added. For example, (1) uses I to refer to the narrator at the text level, but also to both the speaker and the addressee in the inner mental dialogue of the character – even though all these roles are represented by the same subjectivity.

The I in I have to and if I get an early job refers to the character/first person narrator (so it already represents two viewpoints in the text). But the embedded questions (How much should I burden Kristen?) further construct a represented thought pattern and an inner dialogue in which the narrator at the narrative-wide level is also the character/thinker reflecting on future actions, and also the speaker and the addressee in the internal dialogue. This quadruple viewpoint is possible because the questions are embedded in a textually-determined viewpoint hierarchy – from the highest, narrative level, through the intermediate ‘character’s thoughts’ level, down to the faux-direct speech (best understood as ‘demonstration’ here).

In a different structure, described by Vandelanotte (2004, 2009) as Distanced Indirect Speech and Thought, the pronoun I actually replaces the addressee pronoun you in the discourse being reported:

There are the usual signs of Free Indirect Speech here, but the pronoun I reflects a shift of the deictic center to the first-person narrator’s viewpoint. As a result, the reported question in (1), such as How much should I burden Kristen? is formally identical to the questions in (2), such as What did I hope to get from it?, and yet the pronoun I represents the narrator/character, and the inner dialogue participants in (1), while representing the narrator/character and the addressee in a genuine conversation in (2); the speaker in (2) is only profiled contextually, not through the sentence form. In (2), the viewpoint hierarchy thus yields the unusual effect of having I, normally referring to the speaker, refer to the addressee instead.

The pronoun parameter is independent of the remaining formal features, such as the syntactic form of a question, preserved in both examples, or tense, which is different because it is determined by the overall tense choice of the narrative (present in [1] and past in [2]). Each parameter works largely independently of other formal choices, but the meaning is fully determined by the viewpoint hierarchy in the text.

My final example comes from a text by the comedian and feminist Bridget Christie (The Guardian, June 22, 2015), which uses I in a non-genuine way, to mock the viewpoint she targets:

While the I of I am a feminist is genuinely expressive of the writer’s viewpoint, the I of I think is not, as it incorporates a belief about how feminists feel about men, assumed to be held by some members of the public, but not by the writer herself. In other words, the writer presents herself from the perspective of what other people might imagine her to think or say, not what she actually thinks or says.

These examples are not prototypical, but they all represent the RS domain. They embed internal, conversational, or contextually available discourse in a higher (narrative) level of discourse. They use the same deictic form (the pronoun I), but the meaning is determined by a specific pattern of perspectivization in each case. What these examples suggest, then, is that analytical approaches to Reported Speech (and Thought) might consider moving away from the constructional framework, while focusing on the formal and interpretive effects from the perspective of viewpoint hierarchies.

In this context, the approach proposed by S&N opens the door to a theoretical stance focused on the broader intricacies of perspectivization, while at the same time inviting its cross-linguistic verification. It offers a renewed take on one of the most complex phenomena addressed by linguists and stylisticians and it will surely prompt much discussion. I will be watching follow-up conversations with great interest.