Final Report Summary - PERSNAL (Perspective and the Self in Natural Language)
The overarching goal of this project was to articulate a theory of perspective in natural language. A fundamental aspect of the human condition is the ability to adopt another individual’s perspective. I can put myself in Michelle Obama’s shoes (say), by imagining or pretending that I am her, and considering what beliefs, emotions or experiences I might then have. What resources does language put at our disposal in order to describe or report such exercises in empathy? In English, I might say, ‘I’m imagining that I am Michelle Obama and I am under pressure to run for political office’; or with a gerund: ‘I’m imagining being Michelle Obama and being under pressure to run for office.’ Or I could use a conditional sentence, ‘If I were Michelle Obama I would be under pressure to run for office.’ One of the questions posed by the project is how the meaning of these so-called ‘counteridentical’ statements should be modeled formally – a controversial issue in both linguistics and philosophy of language.
The project approached this question using a variety of means. Firstly, experiments were conducted in order to better understand native speakers’ intuitions about the meanings of complex counteridentical statements in English. Prior work in linguistics had found that native speakers differ in their judgments about what these types of sentences mean – a stumbling block that has in the past proven somewhat of an obstacle to making progress with modeling these meanings formally. This component of the project, which was conducted in collaboration with psycholinguist Jeruen Dery, provided some needed clarity on this point, as well as suggesting interesting avenues for future experimental work, and shedding light on methodological issues that arise when conducting experimental work on linguistic meaning.
Secondly, the project brought to bear evidence from languages very different from English, in order to better understand how perspective taking is encoded cross-linguistically. If the ability to put oneself in someone else’s shoes is a universal aspect of human nature, is there also a universal strategy employed by all languages in order to talk about such cases of perspective taking? Conversely, are there areas of cross-linguistic variation? By working with native speakers of languages including Korean, German, and Ewe (a Niger-Congo language spoken in Ghana and Togo), I formulated a novel generalization about the interpretation of counteridentical statements. On the basis of this generalization, I proposed a formal account of how the meanings of counteridenticals are generated (the so-called ‘De Se Generalization’).
The core idea underlying the formal account is this: the mechanism by which counteridentical statements are generated is simply a special case of a more general mechanism that is independently needed in order to account for so-called ‘de se’ interpretations. We can illustrate these interpretations by taking the following simple example. Suppose that Mary, a retired university professor, is going through a pile of papers. She finds an essay written by a student long ago, and, not recognizing the text, says, ‘Whoever wrote this essay is a brilliant student!’ Now, suppose that what we know, but Mary doesn’t, is that she wrote the essay herself many years ago, when she herself was a student – it has simply escaped her memory. Can we report what has taken place by saying, ‘Mary claimed that she was a brilliant student’? Many speakers would say no – or at least, that there is an interpretation of the sentence according to which it is false. In order for the sentence to be true, Mary would have had to have said, ‘I am a brilliant student!’ Thus for the sentence to be true it is not enough for Mary to be talking about herself (qua author of the essay); she must be talking about herself in a first personal way. Such reports, which crucially require a first personal concept of the self, are known as de se reports.
According to the account of counteridenticals and the De Se Generalization that was developed over the course of the project, the formal mechanism that is needed in order to account for de se interpretations is exactly the same as that which is needed to analyze counteridentical statements. This suggests that the notions of perspective and the self are intimately connected in the way that they are encoded in natural language – and furthermore, that at their deepest level languages do not essentially vary in how this connection is manifested. Thus, it seems that – rather surprisingly - natural language treats egocentricity and empathy as two sides of the same coin.
In addition to the reports of these findings that have already been published, I am currently working on a monograph which will explore at greater length their implications for current linguistic and philosophical debates. One of the important dimensions of this work is the idea that the relationship between perspective and the self is replicated across other domains. Take the domain of time, for instance. Just as I can imagine that I am someone other than who I am, I can also imagine that it is a different time from what it actually is – that it is morning and not evening, say, or that it is 1918 and not 2018. In work currently in progress, I show that the De Se Generalization is replicated in the domain of linguistic expressions used to talk about time (such as tense and words like now or tomorrow). I expect this study to naturally lead to a broader investigation of the structure and interpretation of tense and temporal meaning across languages – a project for the future. This project will include a study of the acquisition of temporal meaning by children, particularly by comparing how children interpret verbs that are oriented towards the future (eg hope, expect and want) with those that seem more grounded in the here and now (eg think and believe).
Finally, the findings of the project have implications for neighbouring disciplines of linguistics such as philosophy and psychology. In philosophy, there is currently a lively debate taking place about whether there is anything ‘special’ about de se readings in particular, and thoughts about the self in general. I claim that the cross-linguistic and experimental evidence from counteridenticals provides an argument that there is: contrary to what some philosophers and linguists have argued, language employs a dedicated mechanism for expressing de se readings. This is a striking and rather puzzling fact: given how rare the ‘mistaken identity’ situations are that enable us to pinpoint the role of first personal concepts of the self in de se construal, it seems that natural language could have got on quite well without recourse to such a mechanism. Just why language makes available such resources for expressing de se and perspectival interpretations remains mysterious – but I hope that the findings of the present project will provide some clues to that mystery that can be picked up and built upon by future researchers.
The project approached this question using a variety of means. Firstly, experiments were conducted in order to better understand native speakers’ intuitions about the meanings of complex counteridentical statements in English. Prior work in linguistics had found that native speakers differ in their judgments about what these types of sentences mean – a stumbling block that has in the past proven somewhat of an obstacle to making progress with modeling these meanings formally. This component of the project, which was conducted in collaboration with psycholinguist Jeruen Dery, provided some needed clarity on this point, as well as suggesting interesting avenues for future experimental work, and shedding light on methodological issues that arise when conducting experimental work on linguistic meaning.
Secondly, the project brought to bear evidence from languages very different from English, in order to better understand how perspective taking is encoded cross-linguistically. If the ability to put oneself in someone else’s shoes is a universal aspect of human nature, is there also a universal strategy employed by all languages in order to talk about such cases of perspective taking? Conversely, are there areas of cross-linguistic variation? By working with native speakers of languages including Korean, German, and Ewe (a Niger-Congo language spoken in Ghana and Togo), I formulated a novel generalization about the interpretation of counteridentical statements. On the basis of this generalization, I proposed a formal account of how the meanings of counteridenticals are generated (the so-called ‘De Se Generalization’).
The core idea underlying the formal account is this: the mechanism by which counteridentical statements are generated is simply a special case of a more general mechanism that is independently needed in order to account for so-called ‘de se’ interpretations. We can illustrate these interpretations by taking the following simple example. Suppose that Mary, a retired university professor, is going through a pile of papers. She finds an essay written by a student long ago, and, not recognizing the text, says, ‘Whoever wrote this essay is a brilliant student!’ Now, suppose that what we know, but Mary doesn’t, is that she wrote the essay herself many years ago, when she herself was a student – it has simply escaped her memory. Can we report what has taken place by saying, ‘Mary claimed that she was a brilliant student’? Many speakers would say no – or at least, that there is an interpretation of the sentence according to which it is false. In order for the sentence to be true, Mary would have had to have said, ‘I am a brilliant student!’ Thus for the sentence to be true it is not enough for Mary to be talking about herself (qua author of the essay); she must be talking about herself in a first personal way. Such reports, which crucially require a first personal concept of the self, are known as de se reports.
According to the account of counteridenticals and the De Se Generalization that was developed over the course of the project, the formal mechanism that is needed in order to account for de se interpretations is exactly the same as that which is needed to analyze counteridentical statements. This suggests that the notions of perspective and the self are intimately connected in the way that they are encoded in natural language – and furthermore, that at their deepest level languages do not essentially vary in how this connection is manifested. Thus, it seems that – rather surprisingly - natural language treats egocentricity and empathy as two sides of the same coin.
In addition to the reports of these findings that have already been published, I am currently working on a monograph which will explore at greater length their implications for current linguistic and philosophical debates. One of the important dimensions of this work is the idea that the relationship between perspective and the self is replicated across other domains. Take the domain of time, for instance. Just as I can imagine that I am someone other than who I am, I can also imagine that it is a different time from what it actually is – that it is morning and not evening, say, or that it is 1918 and not 2018. In work currently in progress, I show that the De Se Generalization is replicated in the domain of linguistic expressions used to talk about time (such as tense and words like now or tomorrow). I expect this study to naturally lead to a broader investigation of the structure and interpretation of tense and temporal meaning across languages – a project for the future. This project will include a study of the acquisition of temporal meaning by children, particularly by comparing how children interpret verbs that are oriented towards the future (eg hope, expect and want) with those that seem more grounded in the here and now (eg think and believe).
Finally, the findings of the project have implications for neighbouring disciplines of linguistics such as philosophy and psychology. In philosophy, there is currently a lively debate taking place about whether there is anything ‘special’ about de se readings in particular, and thoughts about the self in general. I claim that the cross-linguistic and experimental evidence from counteridenticals provides an argument that there is: contrary to what some philosophers and linguists have argued, language employs a dedicated mechanism for expressing de se readings. This is a striking and rather puzzling fact: given how rare the ‘mistaken identity’ situations are that enable us to pinpoint the role of first personal concepts of the self in de se construal, it seems that natural language could have got on quite well without recourse to such a mechanism. Just why language makes available such resources for expressing de se and perspectival interpretations remains mysterious – but I hope that the findings of the present project will provide some clues to that mystery that can be picked up and built upon by future researchers.
