1 Introduction

According to Harley (2008) and Nevins (2011), languages that exhibit systematic patterns of morphological syncretism must involve a rule that derives such syncretism as a “deep” property of the grammar. These authors show that, within Distributed Morphology (DM; Halle and Marantz 1993), the relevant rule is Impoverishment (Bonet 1991, Noyer 1992, Halle and Marantz 1994), which, as a context-sensitive operation, deletes feature Fa. in the context of Fβ. In discussing Ljubljana Slovenian, Nevins (2011) posits Impoverishment of the DUAL-number contrasts in the context of feminine gender. However, Nevins’s argument is based only on the relevant morphological paradigms in isolation and only on their nominative case forms. This squib provides more empirical context: namely, entire morphological paradigms from Ljubljana Slovenian, and the interaction of the relevant syncretism with agreement patterns. While the agreement patterns confirm the postsyntactic nature of Impoverishment, the full morphological paradigms show that Impoverishment is systematically blocked in certain case forms: while Impoverishment applies in the context of flexional morphology, it fails to do so in the context of agglutinative morphology. This pattern of blocking can be captured as an intervention effect if Impoverishment is limited to considering a strictly local X0 as context, namely, the closest X0 available in the c-command domain.

While the locality constraints of certain PF phenomena, such as allomorphy, have been the subject of much discussion (Embick 2010, Bobaljik 2012, Merchant 2015, Moskal 2015, Moskal and Smith 2016), the limits of Impoverishment have not been explored to such an extent. The present squib offers to amend this, by proposing that Impoverishment, like many other principles of grammar, obeys strict locality conditions.

2 DUAL-Syncretism

Number contrasts show systematic syncretism in Ljubljana (LJ) Slovenian. Like Standard Slovenian (Toporišič 2000), LJ Slovenian has a three-way gender system (masculine vs. feminine vs. neuter) and a three-way number system (singular vs. dual vs. plural). Gender is encoded in the paradigms of nouns, pronouns, adjectives, and participles.1 Note that, in nouns, one major masculine (M) class and one neuter (N) class exist, while for feminines (F) two major classes exist (F1 and F2 below).

(1)

graphic

(2)

graphic

(3)

graphic

Notice that the DU-PL forms are completely syncretic in the context of feminine gender. Furthermore, it seems that this is not merely a superficial type of syncretism since it is expressed through different exponents. Specifically, the two feminine paradigms in nouns both express this syncretism, but with different exponents, /-e/ and /-i/. Other syncretisms also occur, but I will focus on the number patterns in the context of feminine gender.2 This is the pattern discussed by Nevins (2011), though he only considers nouns.

This same basic pattern of syncretism described by Nevins (2011) for nouns can also be found in the system of pronouns. Consider personal pronouns and proximal and distal demonstratives, for which I supply only feminine forms, but across all three numbers.

(4)

graphic

(5)

graphic

No DUAL contrast3 appears to surface in feminine pronouns.4 Before we look further at this syncretism pattern, let us also consider a verbal paradigm.

(6)

graphic

Verbs express full number contrast: the DU-PL contrast is not syncretic at all. But this is not unexpected given that verbs do not encode gender at all, and the DU-PL syncretisms observed earlier are tied to feminine contexts.

The syncretism of DU-PL forms in the context of feminine gender is pervasive throughout the dialect and, as already discussed, cuts across paradigms in nouns, manifesting through different exponents. According to Harley (2008) and Nevins (2011), such patterns are instances of metasyncretism; these are essentially syncretism patterns that imply a deeper morphological generalization, in this case actual neutralization of number contrast. To capture this, simple Vocabulary item underspecification, as in (7), could be used.

(7)

  • [−SG, +F] ↔ -e / √RTList1 ____

  • [−SG, +F] ↔ -i / √RTList2 ____

However, as Harley (2008) and Nevins (2011) argue, such an account renders the syncretisms completely accidental and misses a morphological generalization. To actually neutralize the contrasts, Harley and Nevins argue for the use of Impoverishment rules. In what follows, I represent number and gender contrasts with a binary feature system, using [±SG, ±AUG] for number and [±M, ±F] for gender (Harbour 2003, Nevins 2011).

(8)

  • Impoverishment in LJ Slovenian (first version)

  • AUG] →

    graphic
    / ___ [+F]

A rule like this must be posited to derive the DU-PL neutralization. The deletion of the [±AUG] feature in the context of feminine gender results in a two-way number system, expressed solely by [±SG]. This is essentially the analysis that Nevins (2011) gives for LJ Slovenian, and it captures the desired generalization.5

3 Agreement Interactions

I now consider further new data, from φ-agreement, which provide more empirical context for the claim that Impoverishment is involved in the feminine paradigms as proposed by Nevins (2011). The agreement data confirm the standard assumption that Impoverishment is a PF/postsyntactic operation. Note that with tensed verbs, the agreement is for number-person but not gender, since verbs never encode gender in LJ Slovenian.

(9)

graphic

These examples illustrate that, although the DU-PL contrast is neutralized on the DP, the respective dual or plural agreement is still reflected on the verb in LJ Slovenian. This is also true of coordinated structures, where two feminine DPs are involved.

(10)

graphic

The agreement pattern in this dialect fits a typical DM analysis perfectly. Agreement relations take place in narrow syntax—or at an early stage of PF (Arregi and Nevins 2012)—where this dialect needs to still express the DU-PL contrasts with all genders. Subsequently, at PF, Impoverishment neutralizes the relevant contrasts. However, this only happens in the context of feminine gender; verbs are not affected, since they do not agree for gender and easily continue to express DU-PL contrasts with feminine subjects even at PF. This is new evidence for Nevins’s proposal of DUAL-Impoverishment in LJ Slovenian. One could easily imagine that the metasyncretisms were in fact so deep that this would be reflected in the lexical feature bundling of the Numeration set, as Harley (2008:292) tentatively speculates. However, the agreement data confirm that the LJ Slovenian neutralization is an active and postsyntactic process.

4 Impoverishment Blocking

A seeming complication arises once entire paradigms of nouns that undergo Impoverishment are examined. In this section, I will pay special attention to the dative and instrumental case forms, which do not seem to undergo Impoverishment at all: I will show that the inflections in these forms are morphologically complex and that this complexity is the reason Impoverishment is absent. Consider all the case forms for the two classes of feminine paradigms, as shown in (11).

(11)

graphic

Syncretism is continued in most nonnominative case forms, but in the dative and instrumental a DU-PL contrast surfaces. In order for Vocabulary Insertion to insert different exponents for these dative and instrumental forms, the DU-PL contrast must be preserved for them.

4.1 Analyses That Modify the Impoverishment Rule

I now consider two analyses of this distribution of Impoverishment, which turn out to be inadequate for several reasons. One option is to make the Impoverishment rule specific to the context of nondative and noninstrumental case forms.

(12)

  • Impoverishment in LJ Slovenian (case-sensitive version)

  • AUG] →

    graphic
    / ____ [+F, {NOM, GEN, ACC, LOC}]

However, such a rule makes incorrect predictions: it predicts that no Impoverishment will occur in contexts where the specified case features are unavailable. This is incorrect since the participles in (3), which encode gender and number but crucially not case, must be subject to the same Impoverishment operation. If Impoverishment were defined as in (12), it would be unable to delete any number features in participles, since the context restriction (the case restriction) could not be met.6

A second option is that two Impoverishment rules are involved: the one in (12) and another one made specific to participles.

(13)

  • Impoverishment in LJ Slovenian (second, participle-specific rule)

  • AUG] →

    graphic
    / _____ [+F]PTC

However, this option misses the generalization that the two rules in (12)–(13) encode the same neutralization effect, just in different word classes.

4.2 Flexion, Agglutination, and Blocking

This now means that there must be another way of deriving the impoverishment in this dialect of Slovenian. A closer look at the case forms where Impoverishment blocking occurs suggests a different solution to the problem. Specifically, I propose that, in the dative and instrumental forms, the gender and number features are not in a sufficiently local relation, which blocks the application of the Impoverishment rule as stated in (8). I propose that these case forms be analyzed as consisting of layers of syntactic heads, and not just a single “fused” head that combines case and all φ-features, as is usually assumed for Slovenian (see, e.g., Börjesson 2006, Caha 2009:240–243).

Independent evidence for this proposal comes from the basic exponent alternations observed in these forms: the noun miza ‘table’ has the datives miz-ama (DU) vs. miz-am (PL), but for kost-

graphic
‘bone’, the datives are kost-ema (DU) vs. kost-em (PL). Also consider the instrumentals: miz-ama (DU) vs. miz-ami (PL), and kost-ema (DU) vs. kost-mi (PL). The same segmental subset of these exponents alternates in the different case-number configurations, which suggests that these should be segmented as follows: miz-a-m-a (DU) vs. miz-a-m-
graphic
(PL), kost-e-m-a (DU) vs. kost-
graphic
-m-i. Further evidence can be found in masculine and neuter paradigms, which reveal more alternations in these forms.

(14)

graphic

Notice that the datives and instrumentals show more alternations—but again, alternations of the same segmental subset of the case-φ “exponent.” Such alternations provide independent evidence that at least three syntactic heads are involved here: a #0 head and a K0 head must be the two final suffixes, while the root-adjacent head is either n0 or a theme suffix.

(15)

graphic

(16)

graphic

In fact, the root-adjacent vowel suffixes cannot be n0 since these suffixes occur above overt nominalizers such as /-its/, as in tigr-its-a ‘tigress (NOM.SG.F)’ ~ tigr-its-a-m-i (INST.PL.F) or žanj-its-a ‘reaper (NOM.SG.F)’ ~ žanj-its-a-m-i (INST.PL.F). These vowel suffixes should probably be characterized as theme suffixes of sorts because they classify nouns according to gender and morphological class: compare the feminine /-a/ and /-e/ in miz-a-m-a and kost-e-m-a in (11), the masculine /-i/ in mošk-i-m-a, and the neuter /-o/ in dreves-o-m-a in (14). In short, n0 is immediately below THM0.7

Notice that the locative suffixes also call for such a decomposition: in (11), miz-a-h (DU/PL) and kost-e-h (DU/PL), and in (14), mošk-i-h and dreves-i-h. /-h/ is likely the spell-out of K0, since it is unique to locative forms, but the preceding vowels could well represent just the theme suffix, or perhaps a combination of the theme suffix with #0.

The observation that dative, instrumental, and locative case forms reveal multimorphemic inflection has been made before for Slavic in general. Halle (1994:46) credits Roman Jakobson with the observation that /m/, among other consonants, appears consistently with all dative and instrumental case forms in Russian, and he posits different classes of theme vowels, similar to what is found in Slovenian. Since the claim that several heads may be merged in the domain of nominal inflection, minimally #0 and K0, is well-established by now (Lamontagne and Travis 1987, Ritter 1992, Travis and Lamontagne 1992), the analysis along the lines of (15)–(16) follows naturally. However, alternative treatments of this phenomenon are possible (Halle 1994), which I address in section 4.3.

A question that must now be answered pertains to the trigger of Impoverishment, namely, gender: where is gender positioned? A significant body of literature argues that gender features occur on n0 (Acquaviva 2008, Kramer 2009, 2015, Deal 2016), and Slovenian also provides evidence for this claim. This assumption follows naturally from the observation that nominalizing suffixes either (a) fix the gender of a genderless stem or (b) change the gender of a stem. As an illustration of (a), consider verbal participial stems constructed with /-l/ (e.g., √hlad-i-l- ‘being cold’) that are further turned into adjectives by the attachment of /-n/ (e.g., √hlad-i-l-n- ‘cold’). This uninflected stem can then be nominalized either by /-its/, which turns our example into the feminine noun hlad-i-l-n-its-a ‘refrigerating room (NOM.SG.F)’, or by /-ik/, which turns it into the masculine noun hlad-i-l-n-ik-

graphic
‘refrigerator (NOM.SG.M)’. (b) can, in turn, be characterized by the following examples:

(17)

graphic

Some roots form bare nouns (with no overt nominalizer) of masculine gender, and some form bare nouns of feminine gender. The attachment of the nominalizer /-its/ to a bare masculine noun then turns it into a feminine one; the attachment of /-k/ to a bare feminine noun turns it into a masculine one. It follows that gender is located on n0 in Slovenian.

The preceding discussion has established that dative and instrumental cases in Slovenian paradigms reveal layers of syntactic heads: n0, a theme suffix, #0, and K0. The remaining case forms systematically show a single exponent of case, φ-features, and THM0 throughout the language, and this suggests that they must undergo fusion of some sort. Whether they are bundled into one head presyntactically through c-selection (Bobaljik 1995, Pylkkänen 2002) or whether this happens postsyntactically is something one can remain agnostic about. However, for convenience, and to be explicit, I will assume that it occurs through the standard application of postsyntactic Fusion (Halle and Marantz 1993:116), which precedes the application of Impoverishment. The nominal paradigms then essentially have a flexional part and an agglutinative part. The relevant Fusion rule can be formalized in the following way, as suggested by an anonymous reviewer:

(18)

  • Fusion rule

  • [[[THM0] #0] K0{NOM, ACC, GEN}] → [THM+#+K{NOM, ACC, GEN}]0

(18) ensures that K0, when specified for NOM, ACC, or GEN, fuses with the immediately preceding head, which bears #-features, and also with the head before that, which contains theme/class features.8 The formalism in (18) also predicts that fusion and agglutination occur in the singular parts of paradigms. This seems to be correct, as some declensions clearly reveal agglutinative morphology in the dative and locative singular forms (e.g., mošk-i-m-u ‘man’ (DAT/LOC.SG.M)) and instrumental forms (e.g., mošk-i-m (INST.SG.M)).

Now that an explicit fusion analysis has been proposed, a new generalization emerges. Notice that the presence of fusion correlates exactly with the presence of Impoverishment, while the absence of Impoverishment correlates exactly with the presence of agglutination, as stated in (19).9

(19)

  • Agglutination-Impoverishment Generalization

  • Agglutination (THM-#-K) correlates with the lack of DUAL Impoverishment.

This is a new generalization that we need to capture. It suggests that a locality constraint on Impoverishment is at play here. I propose that the fusion of THM0, #0, and K0 into one head facilitates Impoverishment, crucially because it renders the number features, [±AUG], local to the gender features contained in n0. In other words, with fusion n0 is the closest X0 that the number features c-command, but in the absence of fusion, the closest X0 that the number features c-command is THM0: THM0 intervenes and blocks Impoverishment. This is essentially a type of Impoverishment intervention effect.10 I state the locality constraint as follows:

(20)

  • Strictly Local Impoverishment

  • Triggering context may be contained in (a) the X0 targeted for Impoverishment, or (b) the closest X0 that the target of Impoverishment c-commands.

This argues against the proposal advanced by Keine (2010), who claims that Impoverishment can never look beyond the contents of the X0 that hosts the target of Impoverishment. A proposal along the lines of (20) is novel in the DM literature and is needed here to explain the pattern of Impoverishment blocking found in the agglutinative parts of the Slovenian nominal paradigm. The only prior evidence for such a locality constraint is reported by Kallulli and Trommer (2011), who propose a similar constraint for Albanian, though for an altogether different kind of pattern, involving nonactive voice morphology; they propose that Impoverishment can “search” for context in the closest head that its target c-commands, and also in the head that immediately c-commands the target. This second option is not encoded in (20), but it could be added. The novel contribution of this squib is a pattern of Impoverishment blocking in the nominal φ-domain, providing evidence for (20).

To see how (20) works, consider a derivation where Impoverishment applies successfully, namely, in nondative and noninstrumental case forms; see (21). In these forms, K0, #0, and THM0 undergo Fusion at PF, which yields a single head. That enables the [±AUG] features to be in the context of the [+F] feature, since [±AUG] immediately c-commands [+F]. Here, the Impoverishment rule from (8) applies.

(21)

graphic

In dative and instrumental case forms, on the other hand, no Fusion applies at PF; see (22). The result of this is agglutinative morphology: here, [±AUG] immediately c-commands only THM0, but crucially not n0, which hosts [+F]. In other words, the closest head in the c-command domain of #0 is THM0. [+F] context is not identified here and Impoverishment is blocked, which leads to the DU-PL contrasts in (11) and (14).

(22)

graphic

This illustrates how the adoption of a locality constraint such as (20) provides a principled account of the distribution of Impoverishment in Slovenian nominal paradigms, and it successfully derives the Agglutination-Impoverishment Generalization stated in (19). In sum, fusion and agglutination conspire to either allow or block Impoverishment.

4.3 Notes on the Multimorphemic Inflection and Fusion

The analysis developed in the previous section gives a principled account of the Agglutination-Impoverishment Generalization (19). Such an analysis provides an argument for the fusion of several syntactic heads in the context of different K0 specifications. This is an important point since the presence and absence of the multimorphemic inflection could in principle be derived by just manipulating the spell-out of different syntactic heads through contextual allomorphy: sometimes they would be overt (in the dative/instrumental/locative forms), but otherwise all except one of them would be null. However, the Impoverishment facts imply that the syntactic structure itself is being manipulated in Slovenian, which calls for the application of a structure-changing operation, namely, Fusion.

However, different analyses of the multimorphemic nominal inflection in Slavic have been proposed without invoking Fusion. Halle (1994:46–51), in particular, is concerned with deriving the presence of theme vowels in Russian. He notes that Russian nouns obey the general word “template” [[[STEM] THM] [#,K]]. He proposes that theme vowels have a certain spell-out, which is then contextually changed by readjustment rules, depending on various φ/K-contexts, but he never proposes that theme vowels undergo Fusion with some other X0. This is justified by empirical facts in Russian (which I do not discuss in detail for reasons of space), which has fully agglutinating sequences throughout adjectival paradigms, and to a lesser extent in nouns. Slovenian is unlike Russian in this respect, since theme vowels are much less pervasive across the nominal and adjectival paradigms; see (11) and (14) for nouns, and (2) and Toporišič 2000:320–329 for adjectives (though cf. footnote 8).

Even so, Halle’s analysis of Russian does not appear be well-suited for the Slovenian situation, since it prevents us from constructing a principled analysis of DUAL Impoverishment.11 We would need to say that THM0 is overt in the dative/instrumental/locative forms, but is spelled out as zero in the remaining forms. In other words, this analysis would imply that the THM0 head has a constant, unfused presence throughout the paradigm and hence cannot be blocking or licensing Impoverishment, and under such an account, we fail to derive the Agglutination-Impoverishment Generalization (19). A principled account of the Impoverishment alternation effectively requires at least THM0 to undergo Fusion to some higher head containing #-features. Such an analysis captures the same generalizations as Halle’s proposal and, in addition, allows us to derive the Impoverishment alternation.

Halle also gives an account of the /m/ found in the dative/instrumental (and /h/ in the locative) in Russian. He proposes that a readjustment rule inserts /m/ between the theme vowel and the final inflection in the context of dative/instrumental features, along the following lines:

(23)

graphic
→ m / THM0 ______ [#, KDAT,INST]0

This also differs from the fusion-based approach proposed here. However, there are two reasons which suggest that the fusion-based approach is on the right track, at least for Slovenian. First, Halle assumes that the inflectional domain in nouns consists of a single morpheme, hence that this must reflect a single syntactic head in DM. However, the claim that several heads may be merged in the domain of nominal inflection, minimally #0 and K0, is well-established by now (Lamontagne and Travis 1987, Ritter 1992, Travis and Lamontagne 1992),12 and there is thus no theoretical reason to reject the idea that the /-m/ in Slovenian is in fact the #0 head.

Second, as illustrated above, at least THM0 must be undergoing some kind of Fusion with the head(s) above it. Since Fusion needs to be used for Slovenian anyway, it seems more parsimonious to analyze /-m/ as the #0 head that participates in this fusion process, and so dispense with the above-mentioned readjustment rules altogether. This analysis, like the one I have proposed for THM0, captures the same generalizations as Halle’s proposal and is more parsimonious in an analysis of Slovenian.13

5 Conclusion

In this squib, I have shown that the DUAL-number neutralization in LJ Slovenian truly is a postsyntactic effect because the neutralization occurs “too late” to affect φ-agreement. I have also discussed a pattern of Impoverishment blocking in subparts of Slovenian nominal paradigms. To derive this blocking pattern, I argued that it needs to follow from a Strict Locality Constraint, where only the contents of the immediately c-commanded X0 can trigger Impoverishment. This indicates that the application of Impoverishment is regulated by very strict conditions, allowing it to emerge as fully principled, rather than being an unconstrained operation at the PF interface.

Notes

1 The data are based on the author’s observation of spoken Slovenian. In addition, all the data points here were checked with two speakers of LJ Slovenian. Note that there is dialectal variation within the area of Ljubljana itself. The data here represent the version of this dialect as observed by the author and as spoken by the individuals who were consulted. All the data points were constructed and elicited by the author, unless noted otherwise.

2 Note that DU-PL syncretism occurs with the neuter forms, as well. These forms will not be discussed here, primarily because it is not fully clear whether they entail actual Impoverishment. See footnote 5 for additional observations on this pattern.

3 Note that the numeral dv-e ‘two’ can be optionally attached to pronouns, to apparently “reinforce” the DUAL contrast. However, since this attachment is optional and since it does not affect the DP triggering DUAL agreement (i.e., these pronouns trigger DUAL agreement just like regular DPs; see section 3), we can conclude that this is not a counterexample to the pervasive DU-PL syncretism identified by Nevins (2011).

4 All masculine pronouns reveal full contrast between DU-PL forms, as expected. This contrast is manifested by the obligatory attachment of dv-a ‘two’ to the pronouns: for example, mi-dv-a (1P.DU) vs. mi (1P.PL). This contrasts with the feminine pronouns, where this attachment is optional.

5 As mentioned in footnote 2, DU-PL syncretism occurs with the neuter forms, as well, which I will not discuss here. It is not fully clear whether the neuter forms are an instance of real Impoverishment. The neuter syncretism does not actually have any counterexamples in this dialect where the DU-PL contrast would arise, but it is true that it is also not a case of metasyncretism, since only one relevant neuter paradigm exists in the language. This leads Nevins (2011:435–439), for instance, to analyze the neuter number pattern as a superficial type of syncretism derived by Vocabulary item underspecification for number. However, a reviewer points out that the rule in (8) could easily be reformulated so as to account for the neuter syncretism, as well: the context specification could be changed from [+F] to [−M]. This is indeed a possibility that I will allow here. However, it is interesting that cross-dialectal considerations suggest that the neuter truly does reflect superficial syncretism. In Božič 2018, I show that a related dialect, Novo mesto Slovenian, is identical to LJ Slovenian in the relevant respects, but differs in that it reveals Impoverishment as an even “deeper” grammatical process; Impoverishment seems to precede φ-agreement in the context of [+F], so that no DUAL agreement occurs on the verb. However, in the masculine and neuter genders, DUAL agreement is transmitted to the verb in the expected way.

6 However, one might wonder whether participles are perhaps never the target of Impoverishment: since they are probing predicates, perhaps they always probe DP goals that have already been impoverished? In principle, such an analysis is possible if we adopt Agree-Link/Copy (Arregi and Nevins 2012), where Agree-Copy values probes postsyntactically at PF. In addition, we would need to say that Agree-Copy occurs after the application of Impoverishment. However, recall from section 3 that tensed verbs systematically show DUAL agreement with DPs that undergo Impoverishment, meaning that Agree-Copy occurs before and not after Impoverishment. We could stipulate that Agree-Copy is exceptionally ordered after Impoverishment just for participles and for no other word class, but this is too costly a stipulation; that is, in the absence of independent evidence, it is a brute-force solution to the problem.

7 This analysis of the multimorphemic inflection can be readily formalized in DM. It involves positing /-m/ as an underspecified Vocabulary item for #0 (perhaps being contextually determined by K0DAT/INST), while the following vowels are exponents of K0 that are sensitive to the DU-PL features in the adjacent #0.

8 The analysis in (18) predicts that all nominative forms in the language will undergo Fusion. However, the inflection found in nominative singular forms in some declensions matches the theme vowel found in the complex multimorphemic forms; compare sonc-e ‘sun’ (NOM.SG.N) ~ sonc-e-m-a (DAT.DU.N) and miz-a ‘table’ (NOM.SG.F) ~ miz-a-m-a (DAT.DU.F). One option is that more heavily specified Fusion rules exist, implying that different K-specifications fuse with only some #-features. But it is more likely that fusion is still occurring in these nominative singular forms, but there is simply no other Vocabulary item available than the one for THM0; there is no independent item for nominative case, and there also seems to be no independent item for #—the closest item for the latter is /-m/, but /-m/ is likely contextually conditioned 0 by dative/instrumental (see footnote 7). In this scenario, THM alone expones [K+#+THM]0, in accordance with the Subset Principle. Since the DU/PL exponents must be targeting fused case-number features, making reference to non-singular specification, they cannot apply in the nominative singular.

9 One could wonder whether the n0, which is often null, is also not subject to Fusion: in nondative/noninstrumental forms it could fuse with [K0+#0+THM0] and in dative/instrumental forms just with THM0. However, when n0 is overt, as in the cases with overt nominalizers, it is always distinct from THM0: consider √rož-

graphic
-a-m-i ‘flowers (INST.PL.F)’, where n0 is null, and √slon-its-a-m-i ‘elephants (INST.PL.F)’, where it is overt. Since the Impoverishment pattern is the same with overt nominalizers, n0 cannot undergo systematic Fusion.

10 Notice that the locative forms also reveal agglutinative morphology, as discussed (cf. miz-a-h ‘tables (LOC.DU/PL.F)’, kost-e-h ‘bones (LOC.DU/PL.F)’), but they never show DU-PL contrasts, which implies the presence of Impoverishment and casts doubt on the generalization stated in (19). However, this is in fact not an issue, as locative forms participate in a broader syncretism/Impoverishment pattern. Consider the masculine, neuter, and feminine paradigms and notice that no DU-PL contrasts ever surface in the locative case forms (recall that masculine gender reveals full DU-PL contrasts otherwise). Whether this is just superficial syncretism or whether it reflects Impoverishment is unclear at this point; if it reflects Impoverishment, no reference to n0GEND is needed, as the trigger is K0LOC and #0±AUG is the target, and these are either “adjacent” or fused into a single X0, which makes for a different syncretism pattern.

11 Note that Russian does not have a DUAL number system and so does not show the discussed pattern.

12 While more recent approaches, such as Nanosyntax (Caha 2009), take this a step further and greatly proliferate the number of heads above n0. See also Leu 2015.

13 A reviewer also suggests that, instead of fusion, fission (Noyer 1992) might be responsible for the multimorphemic inflection in the dative/ instrumental/locative forms. To the extent that fission can be defined in a way that allows it to split a single syntactic head into three heads that occur in a sequence, it could indeed be at play in Slovenian. However, it is unclear to what extent fission makes different predictions from fusion, implying that these two options cannot be teased apart in the discussed dataset. It is, however, clear that fission would need to be able to precede Impoverishment in order to capture the discussed patterns. Future work should focus on distinguishing between fusion and fission in Slovenian.

Acknowledgments

For comments on the research presented here, I would like to thank Lisa Travis, Nico Baier, Tom Leu, Dan Siddiqi, and two anonymous reviewers. In addition, thanks go to the audience at the 2nd Morphology in Montreal-Ottawa-Toronto Meeting (MoMOT 2) at the Université du Québec à Montréal, where the paper was first presented. Any errors are of course my own.

References

Acquaviva,
Paolo
.
2008
.
Lexical plurals
.
Oxford
:
Oxford University Press
.
Arregi,
Karlos
, and
Andrew
Nevins
.
2012
.
Morphotactics: Basque auxiliaries and the structure of Spellout
.
Dordrecht
:
Springer
.
Bobaljik,
Jonathan David
.
1995
.
Morphosyntax: The syntax of verbal inflection
.
Doctoral dissertation, MIT, Cambridge, MA
.
Bobaljik,
Jonathan David
.
2012
.
Universals in comparative morphology: Suppletion, superlatives, and the structure of words
.
Cambridge, MA
:
MIT Press
.
Bonet,
Eulàlia
.
1991
.
Morphology after syntax: Pronominal clitics in Romance
.
Doctoral dissertation, MIT, Cambridge, MA
.
Börjesson,
Kristin
.
2006
. Argument encoding in Slovene: A Distributed Morphology analysis of Slovene noun declension. In
Subanalysis of argument encoding in Distributed Morphology
, ed. by
Gereon
Müller
and
Jochen
Trommer
,
115
130
.
Leipzig
:
Universität Leipzig
.
Božič,
Jurij
.
2018
.
Agreement and morphological neutralization: Data from two Slovenian dialects
.
Ms., McGill University, Montreal
.
Caha,
Pavel
.
2009
.
The nanosyntax of case
.
Doctoral dissertation, University of Trømso
.
Deal,
Amy Rose
.
2016
.
Plural exponence in the Nez Perce DP: A DM analysis
.
Morphology
26
:
313
339
.
Embick,
David
.
2010
.
Localism versus globalism in morphology and phonology
.
Cambridge, MA
:
MIT Press
.
Halle,
Morris
.
1994
. The Russian declension. In
Perspectives in phonology
, ed. by
Jennifer S.
Cole
and
Charles
Kisseberth
,
29
60
.
Stanford, CA
:
CSLI Publications
.
Halle,
Morris
, and
Alec
Marantz
.
1993
. Distributed Morphology and the pieces of inflection. In
The view from Building 20
, ed. by
Kenneth
Hale
and
Samuel Jay
Keyser
,
111
176
.
Cambridge, MA
:
MIT Press
.
Halle,
Morris
, and
Alec
Marantz
.
1994
. Some key features of Distributed Morphology. In
Papers on phonology and morphology
, ed. by
Andrew
Carnie
and
Heidi
Harley
,
257
288
. MIT Working Papers in Linguistics 21.
Cambridge, MA
:
MIT
, MIT Working Papers in Linguistics.
Harbour,
Daniel
.
2003
.
Elements of number theory
.
Doctoral dissertation, MIT, Cambridge, MA
.
Harley,
Heidi
.
2008
. When is syncretism more than a syncretism? In
Phi theory
, ed. by
Daniel
Harbour
,
David
Adger
, and
Susana
Be´jar
,
251
295
.
New York
:
Oxford University Press
.
Kallulli,
Dalina
, and
Jochen
Trommer
.
2011
.
Closest c-command, Agree and Impoverishment: The morphosyntax of non-active voice in Albanian
.
Acta Linguistica Hungarica
58
:
227
296
.
Keine,
Stefan
.
2010
.
Case and agreement from fringe to core: A Minimalist approach
.
Berlin
:
De Gruyter
.
Kramer,
Ruth
.
2009
.
Definite markers, phi-features, and agreement: A morphosyntactic investigation of the Amharic DP
.
Doctoral dissertation, University of California, Santa Cruz
.
Kramer,
Ruth
.
2015
.
The morphosyntax of gender
.
Oxford
:
Oxford University Press
.
Lamontagne,
Greg
, and
Lisa
Travis
.
1987
. The syntax of adjacency. In
Proceedings of the Sixth West Coast Conference on Formal Linguistics
, ed. by
Megan
Crowhurst
,
173
186
.
Stanford, CA
:
Stanford University, Stanford Linguistics Association
.
Leu,
Thomas
.
2015
.
The architecture of determiners
.
New York
:
Oxford University Press
.
Merchant,
Jason
.
2015
.
How much context is enough? Two cases of span-conditioned stem allomorphy
.
Linguistic Inquiry
46
:
273
303
.
Moskal,
Beata
.
2015
.
Domains on the border: Between morphology and phonology
.
Doctoral dissertation, University of Connecticut, Storrs
.
Moskal,
Beata
, and
Peter
W
. Smith.
2016
.
Towards a theory without adjacency: Hyper-contextual VI-rules
.
Morphology
26
:
295
312
.
Nevins,
Andrew
.
2011
.
Marked triggers versus marked targets and impoverishment of the dual
.
Linguistic Inquiry
42
:
413
444
.
Noyer,
Rolf
.
1992
.
Features, positions and affixes in autonomous morphological structure
.
Doctoral dissertation, MIT, Cambridge, MA
.
Pylkkänen,
Liina
.
2002
.
Introducing arguments
.
Doctoral dissertation, MIT, Cambridge, MA
.
Ritter,
Elizabeth
.
1992
.
Cross-linguistic evidence for number phrase
.
Canadian Journal of Linguistics
37
:
197
218
.
Toporišič,
Jože
.
2000
.
Slovenska slovnica
.
Maribor
:
Obzorja
.
Travis,
Lisa
, and
Greg
Lamontagne
.
1992
.
The Case Filter and licensing of empty K
.
Canadian Journal of Linguistics
37
:
157
174
.