1 Introduction

Many languages have large inventories of pronouns, which are licensed in different syntactic and semantic contexts. Pronominal competition arises when multiple pronouns are licit in a given context. This competition often triggers pragmatic effects, such as negative appraisal, emotivity, disambiguation, or register shift (Patel-Grosz and Grosz 2017). Competition between personal and demonstrative pronouns, for instance, can be seen in German and Hebrew. In the German example in (1), personal and demonstrative pronouns are both licensed:Footnote 1

  1. (1)
    figure a

However, use of the demonstrative pronoun—rather than the personal pronoun—triggers negative appraisal of the pronoun’s referent (i.e. an N-effect).

In the recent literature, this type of pronominal competition and the resulting pragmatic effects have been studied exclusively for personal and demonstrative pronouns and in only a small handful of languages. Nevertheless, two classes of analyses emerge to derive these competition-driven pragmatic effects. Referent-based accounts rely on the fact that pronouns often differ in their referential potential (e.g. some pronouns are restricted to discourse subjects, while others are not). Use of a referentially “unexpected” pronoun triggers pragmatic inferences (see e.g. Sichel and Wiltschko 2020). By contrast, structure-based accounts make use of the fact that pronouns can have different amounts of internal structure (e.g. some pronouns contain built-in indices, while others do not). Due to principles of economy, use of a structurally larger pronoun yields a pragmatic effect (see e.g. Patel-Grosz and Grosz 2017).

This paper details a seemingly similar pattern of pronominal competition and negative appraisal in Scottish Gaelic, a Celtic language spoken by approximately 57,400 people (Eberhard et al. 2019). Scottish Gaelic has (at least) three sets of pronouns: personal, emphatic, and (reflexive) anaphoric pronouns, as seen in Table 1.

Table 1 Pronoun inventory of Scottish Gaelic
Full size table

Personal pronouns are the morphological base for emphatic and anaphoric pronouns. Emphatic pronouns are built from personal pronouns with the addition of the suffix -se/-san. This suffix descends historically from the demonstrative seo ‘this’ but no longer carries demonstrative meaning. Anaphors are personal pronouns with an added fhèin particle, which is a derivative of the Indo-European reflexive root *swe-, though the details are complicated and disputed (see e.g. Schrijver 1997).Footnote 2 In many contexts, personal pronouns are ruled out, while emphatic and anaphoric pronouns are both licensed. Precisely in these contexts, use of emphatic pronouns comes across as rude, while anaphors are deemed more polite.

This paper provides a semantics of these three pronoun types and describes and analyzes the N-effect that arises when emphatic pronouns and anaphors are in competition. I argue that personal pronouns are variables that denote individuals in their referential uses, while emphatic and anaphoric pronouns contribute additional meaning. Emphatic pronouns are focus-marked personal pronouns, which invoke individual alternatives to the pronoun’s referent along the lines of Rooth (1992). Anaphors, on the other hand, fall into two classes: plain and exempt. Following Charnavel (2020), I claim that both types of anaphors must be locally bound but by different types of elements; plain anaphors are bound by a local argument DP, while exempt anaphors are bound by a null logophoric pro introduced in the specifier of a logophoric operator. This operator requires its syntactic complement to be evaluated from the first person perspective of prolog’s antecedent.

I show that neither the referent-based, nor structure-based accounts of pronominal competition are able to capture Scottish Gaelic N-effects. Rather, I argue that this negative appraisal is actually the absence of positive appraisal. When a speaker uses an exempt anaphor, they adopt the perspective of the pronoun’s referent, which indicates positive appraisal since they are willing to metaphorically step into the referent’s shoes. Yet when a speaker chooses to use an emphatic pronoun—rather than a similarly licensed exempt anaphor—they refuse to make this perspective shift, which translates as negative appraisal. Consequently, apparent N-effects in Scottish Gaelic are actually not-P-effects.

By providing a semantics of these pronouns in Scottish Gaelic and analyzing a novel form of pronominal competition, this paper contributes to a deeper understanding of pronoun typology cross-linguistically. My analysis also fills a gap in the Celtic literature, where the co-existence of different pronoun paradigms has been noted but not fully explored, and offers a framework that can be used to analyze similar pronouns in other languages. Furthermore, this paper describes and analyzes a new type of pronoun competition. Existing accounts focus on personal and demonstrative pronouns, yet here I show that seemingly similar effects can arise via competition of emphatic pronouns and exempt anaphors. Finally, this paper provides a new way to derive N-effects, which does not rely on reference or structure. Pronoun competition can, then, have superficially similar effects cross-linguistically, but their sources can be quite different.

This paper is structured as follows. Section 2 describes the distribution of the pronouns and offers an analysis of their structure. In Sect. 3, I give a pragmatic analysis of the not-P-effects associated with use of emphatic pronouns rather than exempt anaphors. Section 4 concludes. Data come from my fieldwork on the language based in Stornoway, Isle of Lewis, Scotland. Judgements come from six native Scottish Gaelic speakers living in or around Stornoway. The data primarily reflect Lewis Gaelic, though some speakers originally come from the Isles of Skye, Harris, and Uist.

2 Scottish Gaelic pronoun structure and semantics

2.1 Personal pronouns are phonologically and syntactically small

Data from object postposing and coordination suggest that personal pronouns are phonologically—and, I suggest, structurally—small. First, only personal pronouns can undergo rightward postposing when in object position. Canonically, objects surface immediately after the subject (VSOX). However, personal pronoun objects can shift to the right, optionally surfacing at various points in the clause. Emphatic pronouns and anaphors cannot undergo this rightward shift:

  1. (2)
    figure b

Bennett et al. (2016) analyze an identical pattern of pronoun postposing in Irish, arguing that pronoun displacement occurs to satisfy a phonological constraint. Following them, I assume that the distribution in (2) is phonological at its core and that personal pronouns—but not emphatic and anaphoric pronouns—are phonologically weak.

Second, personal pronouns cannot be coordinated; speakers must use emphatic pronouns and anaphors in coordinations. Although speakers can coordinate emphatic pronouns with anaphors, coordinations containing even one personal pronoun are ungrammatical:

  1. (3)
    figure c

I assume that the pattern in (3) also has a phonological explanation; coordinations in Scottish Gaelic require their conjuncts to be sufficiently phonologically strong, though it is beyond the scope of this paper to formalize this phonological analysis. Rather, I include the postposing and coordination facts to show that personal pronouns are phonologically weak.

Following Cardinaletti and Starke (1999), I translate the phonological weakness of personal pronouns to their syntactic structure, assuming that weak phonology indicates minimal syntactic structure. There are two possible structures that personal pronouns could have. They could be simple ϕPs (4) as in Wiltschko (1998), or they could contain a null NP embedded within a DP shell (5) as in Elbourne (2013).

  1. (4)
    figure d
  1. (5)
    figure e

On either approach, I propose that personal pronouns form the base upon which the other pronoun types are built, since they pattern together for gender mismatch. In Scottish Gaelic, all pronoun types must track the real-world gender of their antecedent, even when this conflicts with its grammatical gender. For instance, boireannach ‘woman’ is grammatically masculine, but—as (6) shows—all pronouns anteceded by this noun must have feminine forms:Footnote 3

  1. (6)
    figure f

Because all three pronoun types pattern together, their base structure must be uniform. For simplicity, I adopt the representation in (4), but my analysis is also compatible with the structure in (5).Footnote 4

2.2 Emphatic pronouns invoke alternatives

Emphatic pronouns invoke alternatives and indicate contrast, which I analyze as focus in the sense of Rooth (1992). Emphatic pronouns are ideal in contexts when a speaker wants to single out an individual from a crowd (7) or highlight contrast between two or more individuals (8):

  1. (7)
    figure g
  1. (8)
    figure h

Furthermore, emphatic pronouns appear with the Scottish Gaelic equivalents of “even” (9) and “only” (10), which are typical association-with-focus contexts (see e.g. van der Wal 2016 for recent discussion):Footnote 5

  1. (9)
    figure i
  1. (10)
    figure j

The structure in (10) is the only way to express at-issue exhaustivity in Scottish Gaelic. This makes it semantically comparable to English “only,” though it is more literally akin to an exceptive construction. Finally, emphatic pronouns are infelicitous in contexts incompatible with focus—for instance, when the referent is established as an aboutness topic as in (11):

  1. (11)
    figure k

Based on these data, I conclude that emphatic pronouns are focus-marked personal pronouns with the schematized structure in (12). The -se/-san suffix realizes a focus head within the nominal domain:

  1. (12)
    figure l

This focus suffix triggers the introduction of a focus semantic value for the pronoun in addition to its preexisting ordinary semantic value (Rooth 1992). The ordinary semantic value is simply the pronoun’s referent as determined by applying the contextual assignment function to the pronoun’s index. The focus semantic value is the set of salient individual alternatives to the pronominal referent. I assume that there is a focus operator (e.g. ∼) higher in the syntax, which operates over the pronoun’s focus semantic value.Footnote 6

2.3 Anaphors come in two flavors

Cross-linguistically anaphors often respect Condition A, which states that they must be bound within their binding domain—roughly the clause (Chomsky 1986). However, anaphors in Scottish Gaelic show two distinct behaviors with respect to this condition. Anaphors can be locally bound, which leads to a reflexive meaning, but they do not need to be—they can simply refer to a salient discourse referent:

  1. (13)
    figure m

Without an appropriate context, the locally bound reading (here indexed with j) is preferred. However, in context, these anaphors can refer to someone else in the discourse. This referent can be the attitude holder of a matrix clause verb (here indexed with i) or another individual who is salient in the discourse (here indexed with k). Consultants report that this discourse-salient individual is often a boss, (ex-)partner, or another common subject of gossipy conversation.

The dual behavior of anaphors in (13) parallels the distinction between plain and exempt anaphora as analyzed by Charnavel (2020). Plain anaphors appear subject to Condition A of binding theory, while exempt anaphors—though morphologically identical to plain anaphors—do not. To unify plain and exempt anaphora, Charnavel proposes that both types of anaphors are subject to strict locality conditions but that they are bound by different elements. An anaphor is plain if it has a local argument DP antecedent. It is exempt if it is bound by a null logophoric pronoun introduced as a specifier of a syntactically represented logophoric operator (Charnavel 2020:215):

  1. (14)
    figure n

The logophoric operator requires that its complement—αP in (14)—be presented from the first person perspective of the null logophoric pronoun (Charnavel 2020: 215):

  1. (15)
    figure o

Antecedents to prolog come in two types: attitude holders who have an intellectual perspective and empathy loci who have a perceptual or emotional perspective.

Charnavel’s analysis accounts for the syntactic and semantic distribution of Scottish Gaelic anaphors. Plain and exempt anaphors have the same syntactic structure—schematized in (16)—and differ only in their binder. I remain agnostic about the syntactic structure exponed by the anaphoric particle fhèin, though I assume that it is indicative of some additional structure (see e.g. Preminger 2019):

  1. (16)
    figure p

Both plain and exempt anaphors are subject to Condition A in Scottish Gaelic, but they appear to differ in this respect due to the nature of their binders. Plain anaphors are anteceded by a local argument DP, whereas exempt anaphors are anteceded by a null prolog introduced by a logophoric operator. This unification of plain and exempt anaphora differentiates Charnavel’s analysis from other approaches to exempt anaphora (e.g. Pollard and Sag 1992), which claim that exempt anaphors simply fall outside the scope of Condition A. Because all anaphors are subject to Condition A on Charnavel’s account, if an anaphor does not have a co-indexed argument antecedent within the binding domain, it must be bound by a null prolog. These exempt anaphors are more limited in their interpretation, since their prolog antecedents are restricted to logophoric centers—specifically attitude holders or empathy loci. The notion of empathy locus explains why Scottish Gaelic speakers consistently report that exempt anaphors are common when speaking about discourse-salient individuals who they have a personal connection with.

The analysis in Charnavel (2020) requires referents of exempt anaphors to be living human individuals capable of serving as perspective holders. Humanness, then, is an important diagnostic of exempt anaphors. The following section discusses humanness restrictions on the antecedents of all three pronoun types in Scottish Gaelic, further confirming the Charnavel-style analysis outlined here.

2.4 Ruling out existing accounts of N-effects

In this section, I show that the existing referent-based and structure-based accounts of N-effects are unable to capture the Scottish Gaelic pattern. Emphatic pronouns and exempt anaphors are both restricted to human referents, making a Sichel and Wiltschko (2020)-style analysis untenable. Likewise, the structural account in Patel-Grosz and Grosz (2017) is unable to derive the specific pragmatic effects associated with emphatic pronouns and exempt anaphors.

Sichel and Wiltschko (2020) derive N-effects in German and Hebrew by taking advantage of the different referential possibilities of personal and demonstrative pronouns. Personal pronouns refer to discourse subjects—entities that participate in the discourse in addition to being talked about. By contrast, demonstrative pronouns pick out discourse objects—entities that can only be talked about.Footnote 7 In this way, a speaker’s choice to use a demonstrative pronoun when a personal pronoun is licensed indicates that the speaker views the referent as someone to be talked about, not with or to, giving rise to negative appraisal. The key difference between personal and demonstrative pronouns stems from their different referential possibilities.

This analysis is not available in Scottish Gaelic because emphatic pronouns and exempt anaphors are referentially restricted in the same way; both pronoun types must have human referents, while personal pronouns can refer to non-humans. This restriction—shared by the two competing pronoun types—makes it unlikely that emphatic pronouns or exempt anaphors refer to different sorts of discourse entities:

  1. (17)
    figure q
  1. (18)
    figure r
  1. (19)
    figure s

Living human referents can antecede all pronoun types (17), yet emphatic pronouns and exempt anaphors cannot be anteceded by animate, non-human (18) or inanimate (19) referents. For exempt anaphors, this restriction falls out of the logophoric analysis; the anaphor’s antecedent must be capable of serving as the perspective holder, which is typically only possible with human referents.Footnote 8 For emphatic pronouns the source of this human-antecedent restriction is less obvious (e.g. perhaps it is an artifact of a larger person-animacy hierarchy that held historically; Griffith 2008). Nevertheless, the key takeaway is that the referents of emphatic pronouns and exempt anaphors are restricted in the same way—they both require human referents. Negative appraisal associated with emphatic pronouns cannot, then, be derived via different referential possibilities of the pronouns, as in Sichel and Wiltschko (2020).

By contrast, Patel-Grosz and Grosz (2017) derive demonstrative pronoun N-effects in German structurally via principles of economy. They argue that pronouns with more syntactic structure require special discourse conditions to be licensed and trigger pragmatic inferences if used when these conditions are not met. Specifically, they claim that German demonstrative pronouns contain a Ddeix projection, which makes them larger than personal pronouns. The use of this larger pronoun leads to N-effects.

However, as Sichel and Wiltschko (2020:46) observe, structural differences alone do not derive a contentful inference; there is nothing a priori negative about using a structurally larger pronoun. I assume that emphatic and anaphoric pronouns in Scottish Gaelic contain the same amount of internal structure. However, even if there were structural differences between these pronoun types, this fact alone would not derive the empirical pattern. For instance, consider the possibility that anaphors are larger than emphatic pronouns.Footnote 9 The analysis in Patel-Grosz and Grosz (2017) predicts that exempt anaphors should trigger negative appraisal, by analogy with structurally large German demonstrative pronouns. Yet this is the exact opposite of the Scottish Gaelic pattern, in which exempt anaphors are associated with positive appraisal. To summarize, the structural analysis in Patel-Grosz and Grosz (2017) is unable to derive the specific pragmatic effects associated with use of a larger pronoun; the derivation of this interpretation must refer to something beyond structure.

3 Deriving N-effects from a lack of perspective taking

3.1 In competition contexts, emphatic pronouns trigger negative appraisal

When emphatic pronouns and exempt anaphors are both licensed, use of emphatic pronouns triggers negative appraisal of the pronoun’s referent, while exempt anaphors come across as more polite. In the contexts in (20)–(22), emphatic pronouns and exempt anaphors are both licensed, but personal pronouns are infelicitous. Precisely in these competition contexts, emphatic pronouns trigger negative appraisal of the pronoun’s referent, while exempt anaphors are associated with positive appraisal (i.e. a P-effect):

  1. (20)
    figure t
  1. (21)
    figure u
  1. (22)
    figure v

Multiple consultants independently report that use of the emphatic pronoun is rude or unfriendly in (20)–(22). Exempt anaphors, on the other hand, signal familiarity and closeness.Footnote 10

Specifically, I assume that the competition seen in (20)–(22) and analyzed in this paper is between emphatic pronouns and focused exempt anaphors, which bear a focus feature but do not contain a syntactically represented foc head.Footnote 11 The semantic contributions of the competing pronouns are, then, only minimally different; emphatic pronouns invoke alternatives, while focused exempt anaphors invoke alternatives and trigger perspective taking.

3.2 Negative appraisal cannot be built into the meaning of emphatic pronouns

Just as with German and Hebrew, emphatic pronouns’ negative appraisal cannot be built into their semantic meaning because it goes away when exempt anaphors are not licensed. In (23)–(24), the speaker does not know the pronoun’s referent, so that individual cannot be an attitude holder or empathy locus referred to by a null prolog.Footnote 12 Since emphatic pronouns require no such perspective taking, they are licensed in the scenarios in (23)–(24):

  1. (23)
    figure w
  1. (24)
    figure x

When exempt anaphors are ruled out, there is no more negative appraisal associated with the referent of the emphatic pronoun, illustrated by the fact that emphatic pronouns are felicitous in sentences expressing overt positive appraisal of the referent. Although it is possible to say positive things about a negatively appraised referent, if the N-effect were encoded in the meaning of the emphatic pronoun, speakers should report a clash between the negative appraisal contributed by the pronoun and the positive appraisal contributed by the utterance as a whole. Even when explicitly asked about this possibility, speakers report no sense of contradiction. In fact, the sentences in (23)–(24) can be followed by even more direct praise of the pronoun’s referent:

  1. (25)
    figure y
  1. (26)
    figure z

The disappearance of N-effects precisely when exempt anaphors are ruled out shows that they must be derived through competition between emphatic and anaphoric pronouns. In this way, the data in (23)–(26) rule out a Potts (2007)-style analysis of emphatic pronouns as expressives. According to Potts, expressive meaning is built into the denotation of expressives, so there is no straightforward way to model competition-driven effects in such a system. Table 2 summarizes felicity judgements for the different pronouns in various contexts. The key observation is that only in contexts of pronominal competition does use of the emphatic pronoun trigger negative appraisal of the pronoun’s referent (cf. row a vs. rows b - g).

Table 2 Summary of pronoun felicity judgements
Full size table

3.3 P-effects arise from perspective taking

Unlike N-effects, I argue that P-effects are not derived through competition, but rather result from the perspective taking triggered by use of exempt anaphors. Following Charnavel (2020), I assume that perspective taking is effectively lexically encoded; although the lexical entries for plain and exempt anaphors are identical themselves, the exempt interpretation necessarily arises when a logophoric operator is introduced into the syntactic structure along with a null prolog in its specifier. The P-effect can, then, be considered a type of expressive content associated with use of the logophoric operator and prolog. Following Potts (2007), I classify this expressive content as a conventional implicature; by using an exempt anaphor, the speaker adopts the perspective of someone else in the discourse, which signals familiarity and closeness with the chosen perspective holder. In this way, the connection between exempt anaphors and P-effects is somewhat indirect—mediated by the logophoric operator and prolog—though these structures necessarily occur alongside exempt anaphors.

Because P-effects are not derived through competition, they should persist in contexts where exempt anaphors do not compete with other pronouns. However, an important hallmark of exempt anaphors cross-linguistically is that they are in free variation with pronouns. This is also the case in Scottish Gaelic, where unfocused exempt anaphors alternate with personal pronouns:Footnote 13

  1. (27)
    figure aa

In (27), emphatic pronouns are ruled out because there is only one relevant individual in the context, while the other pronoun types remain licit. Yet the alternation between personal pronouns and exempt anaphors raises the question of whether it is possible to illustrate P-effects in the absence of any kind of pronominal competition—as needed to show that these effects are not derived through competition.

Despite their shared distribution, I suggest that personal pronouns do not compete with unfocused exempt anaphors. If the personal pronoun competed with the exempt anaphor in (27), my analysis would predict the personal pronoun to trigger N-effects like emphatic pronouns do in focus contexts. Yet this is not the observed pattern; there are simply no appraisal effects associated with the use of personal pronouns. In this way, personal pronouns and unfocused exempt anaphors do not seem to compete with each other. I propose that the absence of competition here has a structural explanation. Because personal pronouns are smaller than exempt anaphors, they act as a default and do not trigger pragmatic effects like appraisal—even when they can alternate with other pronoun types (e.g. exempt anaphors, which remain viable because they contribute additional meaning). This explanation draws inspiration from Patel-Grosz and Grosz (2017), yet—rather than using structure to derive pragmatic effects—I suggest that structural similarity is necessary for pronouns to compete in the first place.Footnote 14

Because personal pronouns and unfocused exempt anaphors do not compete, any P-effects that remain with unfocused exempt anaphors must not be due to competition. P-effects persist, even when emphatic pronouns are ruled out. When the context is controlled to discourage a focus interpretation, exempt anaphors are unnatural in sentences expressing overt negative appraisal:

  1. (28)
    figure ab

Comparison of (27) and (28) shows that, while exempt anaphors are generally licit in contexts barring focus, they become unnatural when expressing overt negative appraisal of the pronoun’s referent. In this way, P-effects associated with exempt anaphors arise even in the absence of competition, suggesting that they are associated with use of the exempt anaphor rather than the failure to choose a different pronoun type.

Further support for the idea that P-effects arise from perspective taking comes from plain anaphors. Plain anaphors in Scottish Gaelic are anteceded by a local argument DP and do not force the proposition to be evaluated from the antecedent’s perspective. Because perspective taking is not necessary for plain anaphors, they can be anteceded by humans (29) and non-humans (30)–(31):

  1. (29)
    figure ac
  1. (30)
    figure ad
  1. (31)
    figure ae

Crucially, plain anaphors do not trigger appraisal effects; speakers can view the antecedents of plain anaphors favorably, though this positive appraisal is not required. In (29), the plain anaphor is anteceded by mo charaid as fhearr ‘my best friend’ which refers to someone who the speaker likes. Yet plain anaphors can also be anteceded by expressions referring to individuals who the speaker expresses overt negativity towards. For instance, it is natural to follow the sentence in (30) with the criticism Abair cù gòrach! ‘What a foolish dog!’. These data show that positive appraisal disappears with plain anaphors, which differ from exempt anaphors only with respect to perspective taking. In this way, there is a demonstrable link between P-effects and perspective taking.

This analysis is also compatible with the fact that exempt anaphors can be used ironically to refer to individuals that the speaker is known to dislike. A younger consultant reports that exempt anaphors are common when talking sarcastically about an ex-partner, often accompanied by an eye roll:

  1. (32)
    figure af

In this use, the speaker subverts the typical P-effects associated with exempt anaphors, since everyone in the conversation knows their true (negative) feelings towards the pronoun’s referent. The use of exempt anaphors is similar to the English expression “my good friend,” which can be used earnestly (33) and sarcastically (34):

  1. (33)
    figure ag
  1. (34)
    figure ah

In fact, Charnavel (2020:167) uses this type of paraphrase to diagnose exempt anaphora, arguing that French cher ‘dear’ identifies empathy loci. To summarize, use of exempt anaphors canonically indicates positive appraisal of the pronoun’s referent, although speakers can play with expectations and use these pronouns sarcastically when talking about someone who they are not on good terms with.

I argue that P-effects associated with exempt anaphors result from the perspective taking that they trigger. This type of appraisal is not specific to competition between emphatic pronouns and exempt anaphors in Scottish Gaelic, so it should not be a language-specific phenomenon. However, to my knowledge, such P-effects have not been documented for exempt anaphors in other languages. I cannot provide a full cross-linguistic analysis here, though I have two suggestions about why P-effects are not yet reported. First, exempt anaphors have a limited syntactic distribution in many languages, which makes studying their pragmatics more challenging. Because exempt anaphors surface in syntactically marked positions and are infrequent in many languages, it is more difficult for speakers to have firm judgements about their appraisal effects. Second, the P-effect triggered by exempt anaphors is subtle, making it easy to miss when not explicitly studying appraisal. P-effects are not the same as politeness in the sense of languages with a T-V distinction; because they arise as conventional implicatures based on perspective taking, P-effects are more varied, registering with some speakers as politeness but with others as friendliness. Although more work is needed, these suggestions offer one way to integrate the Scottish Gaelic pattern into the larger typology of exempt anaphora and appraisal.

3.4 N-effects are actually not-P-effects

As discussed in Sect. 2.4, Scottish Gaelic N-effects cannot be attributed to differences in the pronouns’ referential potential or merely to a difference in size between them. In this way, the existing referent-based and structure-based accounts of pronominal competition cannot capture the Scottish Gaelic pattern.

Rather, I propose that the apparent N-effect arises as a conversational implicature from a speaker’s choice not to use a focused exempt anaphor, which is associated with positive appraisal. Exempt anaphors require the speaker to present the utterance from the first person perspective of the antecedent of prolog. Their use, therefore, signals a willingness on the part of the speaker to take the perspective of the pronoun’s referent, which shows respect, proximity, and friendliness. In contexts where both emphatic and anaphoric pronouns are allowed, a speaker has a choice; they can use either a focused exempt anaphor, which triggers perspective taking and highlights alternatives to the chosen perspective holder, or an emphatic pronoun, which simply invokes alternatives. When a speaker chooses to use an emphatic pronoun rather than an anaphor, there is a conversational implicature that the speaker is not willing to adopt the referent’s perspective. In this way, the apparent N-effect associated with emphatic pronouns is derived from the absence of a P-effect triggered by the use of exempt anaphors.Footnote 15,Footnote 16

Evidence that these N-effects are conversational implicatures comes from the fact that they are reinforceable and cancellable. When the sentences in (20)–(22) contain an emphatic pronoun, they can be followed by an overt reinforcement of the N-effect (a) or a cancellation of this negative appraisal (b):

  1. (35)
    figure ai
  1. (36)
    figure aj

Reinforcing the N-effect triggered by use of the emphatic pronoun is not redundant, and cancelling this negative appraisal is not contradictory. These facts support the claim that N-effects arise via implicature from a speaker’s choice not to use an exempt anaphor.

A consequence of this not-P-effects analysis is that negative appraisal associated with pronouns likely has different sources across languages. In Scottish Gaelic, apparent N-effects exist alongside P-effects. By contrast, Sichel and Wiltschko (2020:28) note that N-effects associated with demonstrative pronouns do not have a positive counterpart in German or Hebrew (or, potentially, cross-linguistically). Even though demonstrative pronouns indicate negative appraisal, personal pronouns do not indicate positive appraisal. Comparison with Scottish Gaelic shows that pronoun competition can have superficially similar outcomes across languages with very different root causes. In fact, my analysis makes a typological prediction. If a language has pronominal P-effects and competition involving this pronoun, then it should also show N-effects, since the choice not to convey proximity or friendliness is inherently negative. However, if a language has pronoun competition that derives N-effects, then it does not necessarily have P-effects because not using a negative pronoun does not obligatorily signify positivity (as in German and Hebrew).

3.5 N-effects pattern differently for 1st vs. 2nd and 3rd person pronouns

This analysis—in which N-effects are derived as not-P-effects—predicts that 1st person pronouns should pattern differently than 2nd and 3rd person pronouns. I assume that sentences are evaluated from the first person perspective of the speaker by default. Given this, use of 1st person exempt anaphors does not trigger a change in perspective taking; they are always evaluated from the speaker’s perspective (at least in a language like Scottish Gaelic without indexical shift). I analyze P-effects as conventional implicatures dependent on this perspective shift, so they cannot arise with 1st person exempt anaphors. Since negative appraisal depends on the availability of P-effects, it also cannot be derived with 1st person pronouns. (37) schematizes this domino effect:

  1. (37)
    figure ak

On the other hand, 2nd and 3rd person pronouns should cluster together, showing the types of not-P-effects analyzed here. Use of 2nd and 3rd person exempt anaphors involves a true perspective shift—from that of the speaker to that of the 2nd or 3rd person referent. This perspective shift generates P-effects, which enables N-effects to be derived for emphatic pronouns:

  1. (38)
    figure al

This prediction about the different behaviors of 1st vs. 2nd and 3rd person pronouns is upheld. 1st person emphatic pronouns and focused exempt anaphors are both common, for example surfacing in response to wh-questions (39) and with “only” (40):Footnote 17

  1. (39)
    figure am
  1. (40)
    figure an

In each of these examples, emphatic and focused exempt anaphoric pronouns are felicitous, making them prime competition contexts. However, there is no evidence of negative appraisal with emphatic pronouns or of positive appraisal with exempt anaphors. Consultants do not report any sense of self-deprecation when the emphatic pronoun is used, nor do they suggest that the anaphor is kinder. The disappearance of these pragmatic effects with 1st person pronouns falls out of my analysis, since they hinge on a perspective shift that does not occur with 1st person pronouns.Footnote 18

2nd person pronouns, on the other hand, show N-effects, since the pronoun-containing propositions are not inherently evaluated from the antecedent’s perspective. Consultants report that 2nd person emphatic pronouns are rude in contexts where focused exempt anaphors are also licensed. In (41) and (42), the emphatic pronoun is inappropriate to say to the referent’s face, since it is unfriendly and abrupt; rather, the speaker should use the exempt anaphor to maintain a good relationship with their addressee:

  1. (41)
    figure ao
  1. (42)
    figure ap

As with 3rd person pronouns, this negative appraisal stems from the speaker’s choice not to adopt the addressee’s perspective, leading via pragmatic reasoning to the conclusion that the speaker is unwilling to step into their addressee’s shoes.

4 Conclusion

This paper described the distribution of personal, emphatic, and (exempt) anaphoric pronouns in Scottish Gaelic and provided an analysis of their semantics. Personal pronouns are variables that denote individuals, while emphatic pronouns invoke focus alternatives and exempt anaphors force the speaker to evaluate the proposition from the first person perspective of the anaphor’s referent. Given these semantics, I offered a pragmatic analysis of competition between emphatic pronouns and focused exempt anaphors, arguing that the negative appraisal associated with emphatic pronouns actually stems from the absence of positive appraisal. In this way, apparent N-effects in Scottish Gaelic are better characterized as not-P-effects. Use of exempt anaphors indicates respect for and proximity with the referent, since the speaker adopts their perspective; when a speaker chooses not to signal this perspective taking by using an emphatic pronoun, apparent negative appraisal results. This analysis shows that perspective taking is relevant in analyzing pronominal competition and its resulting pragmatic effects cross-linguistically. Such a perspective-based account enables the derivation of both P- and N-effects, whose coexistence in Scottish Gaelic highlights a typological asymmetry. N-effects can be basic—as in German and Hebrew—or they can be pragmatically derived—as in Scottish Gaelic. These two types of negative appraisal have superficially similar effects cross-linguistically, but their root cause differs quite substantially.