Small complements of Ps and genitive case assignment

. This paper deals with possessor case assignment in Samburg Izhma-Komi subjects and oblique nominals. The problem is that the case assigned to the possessor depends on the syntactic position of the enclosing nominal; possessors are genitive in subjects and nominative in obliques. This is not predicted by the current theories of case. Looking at morphosyntactic properties of P-complements in Samburg Komi such as possessive agreement within the complement, possessive agreement with the complement, and plural marking of the complement, I show that these complements are not full DPs. The possessor case-marking follows straightforwardly. Since P-complements are not DPs, as opposed to subjects, genitive cannot be assigned. Independently, I show that nominative is the case assigned in PPs. In absence of a D, the case associated with P is assigned to possessors in obliques. This analysis has several theoretical implications. First, I show that inherent cases are better analyzed as syntactic P-heads alongside free-standing postpositions. Second, I provide an argument in favor of DP-hypothesis. Finally, I show that not all arguments within a language must be full DPs.

(1) vas'a-len / *vas'a aj-ys]SUBJ ol-e čomj-yn.In this paper, I provide an analysis of nominal structure in Izhma-Komi with a mechanism of differential case assignment to possessors that avoids locality violations without giving up the locality required by the case theories.Two main ideas form the base of my analysis.First, nonpronominal complements of Ps are not full DPs but rather small nominals.Second, alongside free-standing postpositions, bound morphemes (spatial cases) are also members of the category P.
Small nominals are still an almost unexplored area, but one of their main properties is that they do not require case.Internal case assignment should also be impossible given their size.I will propose the following structure for the oblique nominal in (2).It clearly shows that the possessor DP generated in the Spec,nP can be assigned case from the outside, since small nominals do not include a barrier for locality constraints.This allows for external case-assignment to be local.
Figure 1.The structure of an oblique possessed nominal Since the case in PPs is assigned from outside, while the case in subject DPs is assigned within the DPs themselves, the differential possessor marking is explained, see section 5.
My analysis is based on primary field data collected in the village of Samburg (Temnikovsky district, Yamalo-Nenets Autonomous District, Russia) in the Winter of 2015.Accordingly, the system I analyze here is of the Samburg variety of Izhma-Komi, and it may differ from the other Izhma-Komi varieties.I will refer to the idiom described here as Samburg Izhma-Komi (SIK).In some places where the background is provided and grammatical description seems to correspond to the standard variety (or varieties described elsewhere), I refer to the language as Izhma-Komi.
2. Theoretical background.My analysis relies on four ingredients: (i) the configurational model of case assignment; (ii) P-head analysis of oblique (specifically locative) cases; (iii) hierarchical structure of the noun phrase; and (iv) availability of nominals of different structural sizes within one language.The first three define the framework in which an analysis like the one proposed here can exist.Ingredient (iv) requires more attention.Since I argue for two different nominal categories, I need to provide syntactic diagnostics to detect the difference between them.In this section, I will first discuss (i) through (iii), and then provide the tests for the DP vs. small nominal distinction.
2.1.BASIC ARCHITECTURAL ASSUMPTIONS.I adopt the configurational theory of case (Marantz 1991).Case is assigned according to the disjunctive hierarchy in (3).
(3) Lexically governed > 'Dependent' > Unmarked > Default Lexically governed case is assigned by a head in conjunction with a theta-role.This type of case corresponds to inherent case (Chomsky 1981(Chomsky , 1986)).Since the complements of all postpositions in SIK receive the same marking (or remain unmarked) irrespective of semantics, they are unlikely to assign a lexically governed case.I will not consider a possibility of lexically governed case assignment within PP here.However, I will later discuss the theoretical relation between inherent case and PPs.
Dependent case is assigned to one of two DPs that remain caseless after the assignment of a lexically governed case; these DPs have to stand in an asymmetric c-command relation.There are no multiple DPs in a PP,2 hence no case competition.Thus, dependent case cannot be assigned.
Unmarked case is the next case to be assigned to any DP that has not received lexically governed or dependent case.The choice of the unmarked case depends on the case domain.For instance, nominative is prototypically the unmarked case of the clausal domain, while genitive is prototypically the unmarked case of the nominal domain (Marantz 1991;Baker 2015).In this article, I am concerned with the adpositional domain.This domain is usually not considered in the literature, as opposed to the clausal and nominal domains, Haselbach & Pitteroff (2014) being a notable exception.While Marantz (1991) does not specify what a case domain should be, Baker (2015) defines the domain as the Spell Out of a phase head (e.g.C and D for the clausal and nominal domains respectively).Following the logic, the unmarked case of the adpositional domain is assigned to the only DP in the Spell Out of P (see also Haselbach & Pitteroff 2014).
Finally, default case is used for nominals less integrated in the syntactic structure of a sentence, such as citation forms or hanging topics (Shütze 2001).It is not relevant for the discussion here.
Since I adopt the configurational theory of case, case assignment is independent of Agree (contra Chomsky 2000Chomsky , 2001)).However, I adopt Chomsky's architecture of Agree for morphological agreement.The probe on a functional head looks for a goal bearing specific phi-features in its c-commanding domain.Following Preminger (2014), I assume that the probe can remain unsatisfied and unpronounced if it does not find a suitable goal, and failed Agree does not result in a crashed derivation.
Moving on to the internal structure of PPs, I treat inherent cases as instances of category P, alongside free-standing postpositions (Nikanne 1993;McFadden 2004;Asbury 2008;Dékány 2011;Polinsky 2016, a.o.)On this view, there is no syntactic difference between postpositional phrases and cased forms expressing spatial relations.Both are underlyingly PPs, and inherent cases are expected to pattern with postpositional phrases rather than structural-case markers when there is a difference.For the purposes of this paper, it is important that oblique phrases that are marked with free-standing postpositions and spatial cases exhibit nominative-marking of the possessor uniformly.I will refer to spatial cases as bound Ps.
Finally, I assume that nominals have a hierarchical structure (Abney 1987;Szabolcsi 1984;Bernstein 2001;Alexiadou et al. 2007; Lyutikova 2018, and references therein) consisting of the lexical projection NP and higher functional layers.These include nP in whose specifier the possessor is introduced, as in Figure 1 (Alexiadou et al. 2007), the number projection NumP, and the highest projection DP responsible for integration of the nominal in the higher syntactic structure and discourse: NP < nP < NumP < DP.Some researchers also introduce PossP between NumP and DP, which is responsible for possessor agreement, see (Alexiadou et al. 2007).I do not keep the agreeing projection PossP in the syntactic spine of the nominal, assuming that a phi-feature agreement probe is associated with one of the existing functional heads in the extended projection (see above on the agreement mechanism).I also do not include the case projection KP in the nominal spine (contra Bittner & Hale 1996;Levin 2015;Barrie et al. 2021, who place it on top of DP).Instead, I treat it as a dissociated morpheme (see Embick 1997) introduced post-syntactically.
2.2.SMALL NOMINALS AND THEIR MORPHOSYNTACTIC CHARACTERISTICS.Crucial for my analysis is the expectation that not every nominal within a language must be a full DP (Cinque 2002;Déchaine & Wiltschko 2002, a.o.).Some can be small nominals (terminology from Pereltsvaig 2006).Small nominals are structures smaller than a DP; their reduced syntactic structure is manifested by their morphosyntactic properties, which set them apart from full-fledged DPs.Small nominals usually cannot move, do not get case, have restriction on modification, may have restriction on attachment of certain nominal morphological markers (e.g number, possession, definiteness), and may not be interpreted referentially (Pereltsvaig 2006;Lyutikova & Pereltsvaig 2015). 3 Pleshak (2021) proposes three main diagnostics that distinguish small nominals from full DPs in the complement of a postposition based on data from Moksha, a related Finno-Ugric language.These include (i) possibility of possessive agreement within the complement; (ii) agreement with a non-pronominal complement, and (iii) plural marking of the complement.
In Moksha, if the complement of a P includes a possessor (e.g., behind Pat's house), agreement with that possessor can in principle be marked either on the complement itself (house in the example here) or on the P (behind in our example).According to Pleshak (2021), if possessive agreement is marked on the P-head, that is an indication that the complement of P is a small nominal.Consider (4).The possessive-agreement marker appears on the P-head langə 'on', whose complement is mon' morkš 'my table'.The possessive-agreement marker tracks the phifeatures of the possessor of the complement: mon' 'my'.5).The agreement with the possessor is on the head noun of the complement itself, which is also marked for number and case, as opposed to the head noun in (4).Moreover, the agreement in this case cannot appear on the P-head.In contrast to possessive agreement with pronouns, agreement with non-pronominal goals is optional in Moksha.Hence, langə 'on' bears no possessive marking in (5).
(5) Moksha [mon' morkš-əz'ə-n'] lank-s / *langə-zə-n5 I.GEN table-1SG.POSS.SG-GEN on-ILL on-ILL-1SG.POSS 'on my table ' We have examined possessor-agreement marking in PPs.Now let us consider agreement between relational nouns (a subtype of Ps in Moksha) and their complements.While nominals marked with the genitive together with the definiteness marker trigger agreement (at least optionally) on relational nouns (which served as sources for spatial postpositions and mostly look the same) (6a), bare nominals cannot trigger agreement (6b).( 6 The third diagnostic provided by Pleshak (2021) has to do with plural marking.If a complement cannot be marked as plural, this complement is a small nominal.Example (7) shows that plural marking can be expressed only as long as there is case-marking.However, case-marked forms are DPs, and as such, trigger possessive agreement on the head (7a).
Thus, we see that in contexts where a DP receives case, carries possessive agreement, triggers agreement on the P, and has plural marking, none of these properties are observed with small nominals.
It is worth noting that these tests should be performed with caution.First, if the language does not have agreement in PPs in any context, and there are no minimal pairs with and without agreement to compare, the absence of agreement on its own does not indicate that nominals in PPs are small nominals.It is also not true that having possessive agreement or number marking affixes on the nominal will automatically exclude it being a small nominal (see section 4.1 for an example).
agreement with the possessor agreement on P obligatory overt expression of plurality DP on the head nominal yes yes6 Small nominal on the P-head no no Table 1.Distinguishing DP from small nominals within a PP In the next section, I will show how these tests can be applied to SIK data.

SIK nominal morphosyntax.
In order to discuss the differential possessor marking in the PP, I will first present the basics of SIK nominal morphosyntax.
3.1.SIK NOMINAL MORPHOLOGY.Izhma-Komi has 16 morphological case markers (Biryuk et al. 2010).A part of the case paradigm is represented in When it comes to expression of possession, possessive markers precede some case markers (10a) and follow others (10b).Possessive markers follow those exponents that should be analyzed as bound P-heads, not as case markers.Possessive markers can also be fused with some case markers or P-heads, e.g., accusative in (8) or essive in (10c); see Simonenko & Leontjev (2012) for the distribution of cases with respect to affix order and for an analysis.
(10) a. kerka-je-len b. kerka-tɨ-e c. kerka-am house-POSS.1SG-GENhouse-PROL-POSS.1SGhouse-ESS.POSS.1SG'my house' 'through my house' 'in my house' Possessive agreement appears on the head nouns of possessive noun phrases, on postpositions in PPs, and as subject agreement on non-finite forms.I will describe the morphosyntax of possessive nominals and PPs; possessive agreement on non-finite forms is irrelevant to the discussion here.
The third person possessive marker in Izhma-Komi can also appear on discourse-given entities, thus serving as a definite article (Kashkin 2008;Simonenko 2014).To exclude such instances, I will concentrate on examples with a first-or second-person possessor.

SIK POSSESSIVE NOUN PHRASES.
In possessive phrases, a number of factors determine whether or not possessive agreement has to be marked.With pronominal possessors, possessive agreement is obligatory irrespective of the relation or syntactic position (11), see (Pleshak 2020) for discussion.If the possessor is non-pronominal, obligatoriness of agreement depends on the syntactic position.For instance, in oblique phrases, possessive agreement with non-pronominal possessors is always optional (12).If the phrase is in the subject position, agreement is obligatory for inalienable possession, and optional for alienable possession,8 compare ( 13) and ( 14).These facts are summarized in Figure 2. It follows from this discussion that SIK PPs exhibit possessive agreement, and there is a plural marker to mark number.Therefore, the diagnostics presented in 2.2 should be applicable to SIK.Pronominal possessors are the best diagnostic context for agreement because they require that agreement appear on the head of the construction.

Complements of Ps in SIK.
Using the diagnostics presented in section 2.2, I will now show that non-pronominal complements in SIK are small nominals.I will start this discussion with pronominal complements, which are DPs.This way I will have a baseline for the comparison of other nominals to personal pronouns.Then I turn to the non-pronominal complements themselves.I will discuss the agreement with the possessor of the complement and then consider the agreement with the complement on the P-head as well as the number marking of that complement.I will first present all these observations for free standing P heads, and then separately consider bound Ps.

COMPLEMENTS OF FREE-STANDING POSTPOSITIONS.
If the complement of a postposition is a personal pronoun, then the postposition must agree with it.This is consistent with the status of pronominal complements as full DPs.It is also worth noting that these pronominal DPs do not show any case morphology.Unlike Moksha, pronominal complements in SIK cannot be in the genitive case.Since full DPs are not expected to be caseless, I assume that pronominal complements are assigned nominative in SIK.The fact that case-marked nominals in PPs are nominative, not genitive or marked with some other case, is going to be important for my analysis of possessor case-marking in oblique contexts, see section 5 for more details.
When non-pronominal complements of SIK postpositions are alienably possessed (e.g., 'house' in ( 16)), possessive agreement never appears on the head noun of the complement itself, irrespective of case-marking of the possessor (16b).It surfaces obligatorily on the postposition (16a).With inalienably possessed nouns, illustrated with 'daughter' in (17), there is variation.Possessive marker can appear either on the head noun (17a) or on the postposition (17b).( 16 2021) diagnostic, since P-head is the only locus of possessive agreement in the PP, such non-pronominal complements of postpositions are never full DPs in SIK; they have no structural space for a probe, and they do not prevent an external probe from reaching the possessor, see Figure 3 in section 5 for more detailed analysis.
This diagnostic does not work for inalienably possessed nominals, because they are able to bear possessive suffixes.Since the diagnostic is unidirectional, it does not tell us anything about the categorical status of the nominals that combine with possessive morphology.Two analytical possibilities can be entertained with respect to inalienably possessed P-complements.One option is that such inalienably possessed nominal complements must be DPs.Alternatively, in the presence of possessive marking, such possessed expressions are small nominals, and the possessive probe is simply lower in case of inalienable possession than it is when possession is alienable.(I leave the specific details of the placement of the probe for future research.)I now turn to the second diagnostic of small nominals, which will allow us to evaluate these analytical options.
The second diagnostic concerns the agreement of the postposition with the non-pronominal complement.Crucially, the non-pronominal complement cannot trigger possessive agreement on the postposition even if it is an inalienably possessed nominal bearing possessive agreement itself: The inability of a non-pronominal complement to trigger possessive agreement on the head suggests that this complement is not a full DP.This diagnostic helps us distinguish between the two analytical possibilities regarding the status of inalienably possessed nominals.They must be small nominals, not DPs.
Finally, let us consider the plural marking of the complement.As shown in ( 19), plural marking is at least highly degraded if not ruled out completely.Moreover, based on the fact that plural interpretation of ( 19) is available, the absence of overt morphological marking reflects number neutrality rather than presence of a null singular marker.This result is not definitive.It is unclear exactly to what extent plural marking is ungrammatical and what contributes to its degraded status.However, the plural-marking diagnostic is unidirectional.If there is no plural marking, then the nominal is a small nominal.At the same time, the availability of plural marking does not rule out the option that the P-complement in question is still a small nominal.It may be structurally larger than an nP, but not a DP.On this approach, there is room for the number projection in the nominal spine.More data are needed to tell whether such nominals are smaller than NumP, like in Moksha, or rather NumPs.The first two tests for small nominalhood unambiguously confirm that SIK non-pronominal complements are not full DPs.The final diagnostic is less definitive, but it does not contradict the other two.I therefore conclude that non-pronominal P-complements in SIK are small nominals.The presence of a possessor in the structure leads us to the conclusion that SIK non-pronominal complements are at least nPs.These non-pronominal complements do not receive case, and the head noun surfaces in its bare form.4.2.COMPLEMENTS OF BOUND PS.So far, I have been looking at constructions with free-standing postpositions even though the claim I made formulating the problem was more general: subject nominals vs. oblique nominals, not cased nominals vs. nominals in PPs.The example of an oblique nominal I gave contained a case-marked noun phrase rather than a construction with a free-standing postposition.Before I proceed to my analysis of case-marking of possessors, I would like to show that the same pattern as for free-standing postpositions holds for bound Ps, traditionally treated as case-markers.Remember that I think of case markers as exponents of Pheads.Therefore, the complement of P in this case is the case-marked noun phrase (with exclusion of the case itself).
When it comes to the place of the agreement marker, the POSS+CASE order would correspond to the agreement on the head noun of the complement of a free-standing postposition.The order CASE+POSS, on the other hand, would correspond to the agreement on the postposition.
The fact that, in the contemporary SIK, spatial cases are expressed together with possession within one morphological marker slightly complicates the situation.However, if we take a more attentive look at the fused affixes, we will see that the possession part is at the very end of the affix.Consider a fragment of the possessive paradigm in Table 3.It contains possessive markers in the nominative (which are presumably just possessive markers since nominative is morphologically unmarked) and essive/illative markers.The last part of the essive/illative marker repeats the nominative possessive marker, which is most prominent in plural forms.Based on the considerations above, I conclude that possessive agreement with the possessor of the complement is always on the P-head.
Since possessive marking only occurs in contexts where an overt possessor is present, I conclude that non-pronominal complements (nominal stems) cannot trigger possessive agreement on the bound P (case marker).Personal pronouns, on the other hand, always contain a possessive marker in their case paradigm.Translating this into my analysis, personal pronouns trigger obligatory agreement on the bound P. Consider (20) for elative forms.
(20) a. me-s'um b. mi-s'unum I-EL.POSS.1SGwe-POSS.1PL'from me' 'from us' The only diagnostic that gives a different result for bound P-heads compared to free-standing Pheads in SIK is the possibility of plural marking.The plural marker is compatible with bound Pheads and precedes them, as shown in ( 21).
(21) samburg kerka-jas-yn Samburg house-PL-ESS 'in the houses of Samburg' Recall that this diagnostic is unidirectional.If there is no plural marking available, then the nominal is a small nominal.If plural marking is possible, this does not exclude the option that the nominal is still smaller than a DP.The judgments also were not clear for ( 19) with a free-standing postposition.In any case, bound and free-standing P-heads have slightly different properties in SIK; I leave these more nuanced differences for future research.Nevertheless, there is no counterevidence to the assumption that complements of bound Ps are small nominals, which allows me to proceed with the analysis of possessor case-marking in oblique contexts.

Case-marking in SIK PPs.
The main takeaway so far is that the nominals in the complement of P-heads in SIK cannot be full DPs unless they are personal pronouns.We are now in a position to turn to case assignment.Recall the initial problem presented here, that is, that caseassignment within a noun phrase appears to be sensitive to the external syntactic context, as shown in (1)-( 2) above.This is not predicted under locality of case assignment.Recall also that all DPs must receive case.At the same time, only DPs but not small nominals require case.
My analysis in this section is going to be based on the idea that if different syntactic heads select for nominals of different internal structure, the sensitivity of possessor-marking to the syntactic context external to the enclosing nominal can be easily explained.
Adopting the configurational theory of case, genitive is the unmarked case in the nominal domain, which is the Spell Out of D. In other words, any DP within another DP will be marked with the genitive case (unless it can receive an inherent case).This is exactly what happens in subject DPs, where possessors surface as genitive: In oblique phrases, and in spatial contexts considered here specifically, there is no DP layer, as I argued in section 4. Therefore, genitive cannot be assigned.Nevertheless, we see overt possessors in such contexts, which means that such DPs are licensed and therefore will be assigned some sort of case.The source of this case should be some higher domain.By analogy with other, better explored, case domains, I assume that such a higher domain is the PP domain.
Thus, I propose to abandon the conception that P-heads invariably assign inherent cases and instead introduce a distinct case domain inside a PP, on par with clausal and DP-internal case domains.This move is not a mere stipulation.We see that nominative is the only case assigned by postpositions in SIK, independent of the theta-role, and it is the case assigned to the only DP in the domain (see section 4.1 for a discussion of the case assigned by SIK postpositions, example (15) in particular).Therefore, nominative is the unmarked case of the P-domain, rather than dependent case.Back to the data presented in section 4; recall that non-pronominal complements of Ps are small nominals, not full DPs.Combining this result with the observation that possessor expressions are not in the nominal domain in SIK, we can conclude that possessors of non-pronominal complements are actually the only DPs within the P-domain.As such, they should receive the nominative case.This prediction is borne out.The structure for example ( 23) is given in Figure 3. Here, the complement of P ('in front') is the case domain; in the absence of the DP layer in the complement, the possessor DP ('my') is the only DP in the Spell Out of that P.This possessor DP is visible to PossP, the agreeing probe associated with the P-head, because there are no phase heads intervening between the two.In a similar way, we can account for possessor marking in nominal phrases bearing spatial cases, which are bound Ps.Recall that bound Ps allow for plural marking.I assume that this class of P-heads selects for NumP rather than nP.Although additional evidence for that conclusion is still outstanding.Except for the amount of structure within the small nominal and the phonological form of the P-exponent, the structure for (24), shown in Figure 4, is basically the same as for (23).Again, the nominative marking is expected because the possessor DP is the only DP in the P domain.presentational reasons, as it hosts morphological markers. 10Since PossP is a probe associated with P, and the marker that surfaces represents a fusion of P and possession, the phonological forms on the terminal nodes are rather schematic.I omit the precise mechanism of mapping in the interest of space.
Thus, if we adopt the view that some nominals can be smaller than full DPs and show that such nominals can be selected by P heads, the case marking pattern observed in SIK follows straightforwardly.

Conclusions.
In this paper, I addressed the question of differential possessor marking in SIK, where the case-marking of the possessor seems to depend on the syntactic position of the enclosing nominal.Specifically, the possessor of nominative subjects gets genitive, while possessors of oblique nominals appear in the nominative case.I argued that the difference in possessor marking comes from the difference in the structural size of the possessed nominals.In contrast to the conception that languages can be divided into DP-and NP-type (more on that below), I contend that nominals of different structural size can co-occur within one and the same language.Structural size of a nominal is relativized to the syntactic position in which it occurs.In SIK, the subject position has to be occupied by DPs, while the complement of P must be a small nominal, either an nP or a NumP.
I showed that the size of the nominal in SIK has consequences for the case marking of possessor within that nominal, a novel result in the analysis of noun phrases and PPs.In the absence of the D head in a nominal projection, the nominal case-assigning domain cannot be created, and the possessor gets assigned case within the closest local available domain.This domain is the postpositional domain, whose unmarked case in SIK is nominative.
The theoretical implications of these results are twofold.The first set of implications has to do with the status of inherent cases in the grammar.Unless one adopts the view that inherent cases are structurally Ps, it is problematic to explain why nominals marked with inherent cases pattern with PPs with free-standing postpositions, rather than with other case-marked nominals.
The second set of implications concerns the distribution of nominals of different sizes across languages.The assumption that nominals, at least in some languages, must be DPs rather than NPs has been mostly taken for granted, and the argumentation provided so far is mostly based on outdated theory-internal assumptions and the desired parallelism between nominal and clausal structure.Salzmann (2020) discusses the arguments for the DP hypothesis (as well as the arguments for the NP hypothesis) in detail and shows that they are rather inconclusive.He also notices that a stronger argument in favor of the presence of D would feature syntactic heads that do not select for DPs and only select for smaller nominals.As I showed, Komi Ps are of the needed type; they can only take non-pronominal complements that are smaller than full DPs.This would be difficult to explain under the NP hypothesis, which denies higher functional projections on top of NP (see Bruening et al. 2018 andBruening 2020 for arguments pro the NP hypothesis).Therefore, my analysis provides a valid argument in favor of the DP hypothesis.
Second, those researchers that adopt the DP hypothesis, assuming that D should be the actual head of a fully-projected nominal, hold different points of view.One point of view supports the universal DP hypothesis, meaning that all nominals in all languages must be DPs, regardless of the presence of overt D-elements, such as articles (Longobardi 1994).Another point of view is the parametrized DP hypothesis, claiming that only languages with articles project DP, while articleless languages only have NPs (Bošković 2008).According to Bošković, the NP/DP parameter correlates with various other characteristics of the language, such as possibility of Left Branch Extraction and licensing of NPIs under negative raising.These concomitant characteristics can be used as diagnostics to determine whether a language is of the NP-or DP-type.
Most papers arguing against the parametrized DP hypothesis focused on the fact that Bošković's diagnostics are inconclusive and cannot be used to determine the NP/DP status of nominal arguments (see Köylü 2021 for an overview).A stronger argument against Bošković's intra-linguistical parametrization is evidence that nominals of different size can co-occur within one language.Several researchers have offered such evidence (Déchaine & Wiltschko 2002, Pereltsvaig 2006;Lyutikova & Pereltsvaig 2015;Erschler 2019), and this paper adds data from SIK, with its structural-size asymmetry between subjects and P-complements, to the list.

Figure 4 .
Figure 4.The structure for a PP with a possessed NumP The head noun of the complement morkš 'table' has no agreement marker and appears bare.

Table 2
Even though inherent cases including dative and below are traditionally considered part of the case paradigm, I consider them to be exponents of P-headsbound Ps, see section 2.1.Expression of number is relatively straightforward.For plural nominals, the affix -jas is added before the case marker.
. Nominative does not have an overt case marker.There are two genitive cases; the second genitive only marks possessors of accusative objects.7Accusativecase only occurs in combination with a possessive marker (8); direct objects, which do not receive possessive markers, remain bare.