Morphosyntactic alignment
Morphosyntactic alignment refers to the systematic ways in which languages encode the grammatical relations among the core arguments of verbs—the single argument (S) of intransitive clauses, the agent-like argument (A) of transitive clauses, and the patient-like argument (P)—through morphological markers such as case affixes or adpositions, verbal agreement, or syntactic positioning.[1] This alignment determines how these arguments "behave alike or differently" in terms of grammatical properties, reflecting the linkage between thematic roles (e.g., agent, patient) and syntactic functions (e.g., subject, object).[2] The concept is central to linguistic typology, as it varies across languages and influences how semantic roles are mapped onto grammatical structures.[3] The primary types of morphosyntactic alignment include nominative-accusative, ergative-absolutive, and others, each characterized by distinct patterns of grouping the arguments S, A, and P. In nominative-accusative alignment, the most widespread pattern (occurring in approximately 32% of languages), S and A are treated similarly (e.g., both marked as nominative), while P receives distinct marking (e.g., accusative), as seen in languages like English and Spanish.[1] Conversely, ergative-absolutive alignment (about 14% of languages) groups S and P together (absolutive), distinguishing A (ergative), a pattern evident in languages such as Basque and many Australian Aboriginal languages.[1][3] Neutral alignment, where all three arguments (S, A, P) receive identical or no marking, is also common (around 49% of languages), as in Mandarin Chinese, while rarer types include tripartite alignment (S, A, P all distinct, ~2%) and active-inactive (split-S based on agentivity, ~2%).[1][3] Many languages exhibit split alignments, where the pattern varies according to factors like tense-aspect-mood (e.g., ergative in past tenses), person/animacy hierarchies (e.g., accusative for pronouns but ergative for nouns), or semantic verb classes, leading to hybrid systems that complicate typological classification.[2] Diachronically, shifts in alignment—such as from ergative to accusative—often arise through reanalysis of low-transitivity clauses or pragmatic pressures, though such changes are rare and constrained by learnability and structural factors.[2] These variations highlight morphosyntactic alignment's role in understanding language diversity, universals, and evolution, with cross-linguistic databases like the World Atlas of Language Structures providing empirical foundations for comparative analysis.[3]Fundamentals
Definition and Scope
Morphosyntactic alignment refers to the ways in which languages pattern grammatical markers across the core arguments of clauses, specifically grouping and marking the single argument (S) of an intransitive verb with either the agent-like argument (A) or the patient-like argument (P) of a transitive verb through mechanisms such as case marking, verb agreement, or word order.[4] This concept, formalized in typological linguistics using Dixon's (1979, 1994) terminology for S, A, and P, captures how languages encode syntactic relations without directly reflecting semantic roles like agent or patient, though semantic factors can influence alignment patterns.[3] The scope of morphosyntactic alignment extends to both morphological and syntactic domains: morphologically, it involves case affixes or clitics on nouns and pronouns; syntactically, it manifests in verb agreement with specific arguments or the selection of syntactic pivots for complex constructions like relativization.[5] Unlike semantic role labeling, which prioritizes thematic relations (e.g., who initiates an action), alignment focuses on consistent grammatical treatment of arguments across clause types, allowing for cross-linguistic comparisons of how S aligns with A or P.[4] This distinction ensures that alignment analysis remains centered on formal encoding rather than event semantics. To illustrate basic clause structures, consider English, which exemplifies nominative-accusative alignment: in the intransitive clause "The dog runs," S ("the dog") receives nominative marking (or is unmarked for full NPs) and patterns with A in the transitive "The dog chases the cat," where P ("the cat") is accusative (often unmarked for NPs but distinct in pronouns, e.g., "He chases him").[3] In contrast, Basque displays ergative-absolutive alignment: the intransitive "Mutila etorri da" ("The boy arrived") marks S ("mutila," absolutive) identically to P in the transitive "Gizonak mutila ikusi du*" ("The man saw the boy," with "gizonak" ergative for A and "mutila" absolutive for P), highlighting how S and P share unmarked absolutive case while A is distinctly marked.[6] Alignment patterns are not always uniform within a language; they may vary by grammatical construction, such as tense-aspect (e.g., ergative in past tenses but accusative in present) or person (e.g., different treatment of first/second vs. third person arguments), allowing for hybrid systems that reflect historical or functional pressures.[7]Historical Development
The concept of morphosyntactic alignment traces its roots to 19th-century linguistic descriptions of non-Indo-European languages, where scholars first documented grammatical patterns diverging from the nominative-accusative systems dominant in European languages. In Australia, missionary grammarians provided early accounts of ergative-absolutive case marking in Pama-Nyungan languages; for instance, Lancelot E. Threlkeld's 1834 grammar of Awabakal distinguished an "Active Nominative" case for transitive agents, while leaving intransitive subjects and transitive patients unmarked or in a default form, highlighting the unified treatment of the latter two roles.[8] Similarly, in the Caucasus region, Russian military officer and linguist Peter K. Uslar produced detailed grammars of Northeast Caucasian languages during the 1860s–1890s, such as his 1889 work on Avar and 1892 description of Lezgi, which systematically outlined ergative case assignment to transitive subjects alongside absolutive marking for intransitive subjects and transitive objects.[9] These pioneering efforts, often framed through Latin-inspired terminology like "nominative" for agents, revealed alignment as a cross-linguistic variable rather than a deviation from a universal norm. Early 20th-century anthropologists and linguists expanded on these observations by exploring alignment variations in indigenous languages of the Americas. Edward Sapir, in his 1917 review of C.C. Uhlenbeck's work on Algonquian languages, analyzed active-stative (or active-inactive) patterns in several Native American languages, arguing that intransitive subjects receive distinct markings based on verb semantics—active agents aligning with transitive subjects, while stative patients align with transitive objects—thus introducing a semantic basis for split systems.[10] This contributed to growing recognition of alignment diversity beyond rigid accusative or ergative molds, influencing structuralist approaches to morphology. The mid-20th century saw the institutionalization of alignment within typological linguistics, spurred by Joseph H. Greenberg's 1963 formulation of grammatical universals, which treated case and agreement patterns—including alignments—as comparable parameters for assessing language universals and implications, such as the tendency for SOV order to correlate with certain marking strategies. This shift from historical-comparative to typological methods in the 1960s–1970s provided tools for cross-linguistic analysis, emphasizing alignment's role in encoding transitivity and argument structure. Modern theoretical formalization of morphosyntactic alignment crystallized in the 1970s through key publications that defined core terms and frameworks. Stephen R. Anderson's 1976 chapter "On the Notion of Subject in Ergative Languages" coined "split ergativity" to characterize systems where alignment shifts contextually (e.g., by tense-aspect or nominal hierarchy), challenging unified notions of "subject" and integrating ergative patterns into syntactic theory.[11] Building on this, R.M.W. Dixon's 1979 article "Ergativity," published in Language, offered a comprehensive typology of ergative systems, distinguishing morphological from syntactic ergativity and positioning alignment as a fundamental axis for grammatical comparison, with implications for universals and language change. These milestones elevated alignment from descriptive anomaly to a cornerstone of linguistic typology.Core Alignment Types
Nominative-Accusative Alignment
Nominative-accusative alignment is a morphosyntactic pattern in which the single argument of an intransitive verb (S) and the agent-like argument of a transitive verb (A) receive the same grammatical treatment, distinct from the patient-like argument of a transitive verb (P). In this system, S and A are typically marked as nominative—often unmarked or with a dedicated case marker—while P is marked as accusative, either overtly or through differential coding.[3] This alignment organizes core arguments into two categories: a unified subject category (S/A) that functions as the pivot in syntactic constructions, and an object category (P) that is treated separately.[3] The pattern manifests through various grammatical mechanisms. Morphologically, it appears in case affixes, as in Latin where the nominative marks S and A (e.g., puella "girl-NOM" in both Puella currit "The girl runs" and Puella felum videt "The girl sees the cat"), while the accusative marks P (felum "cat-ACC").[3] In verb agreement, languages like Italian exhibit nominative-accusative alignment by having finite verbs agree in person and number with S or A but not P; for instance, Io parlo "I speak" (intransitive, verb agrees with S io) parallels Io parlo italiano "I speak Italian" (transitive, verb agrees with A io), with no agreement on P italiano. Syntactically and pragmatically, English relies on fixed subject-pivot word order, where S/A precede the verb as the privileged syntactic role (e.g., "The dog chased the cat," with "the dog" as S/A in subject position and "the cat" as P in object position), facilitating constructions like subject extraction in relatives.[12] This alignment is the most prevalent cross-linguistically, occurring in approximately 55% of sampled languages for verbal person marking.[12] It dominates major language families, including Indo-European (e.g., English, Hindi), where it structures case and agreement systems; Austronesian, where the majority of languages exhibit nominative-accusative patterns in argument coding; and Niger-Congo, as seen in Bantu and Kwa languages like Avatime, which select S/A as the primary syntactic argument.[13][14] Subtypes of nominative-accusative alignment vary by domain. Morphological subtypes involve explicit affixes on nouns or pronouns to distinguish S/A from P, as in Hungarian's nominative zero-marking versus accusative -t.[3] Syntactic subtypes privilege S/A in clause-level behaviors, such as relative clause formation, where subjects are most accessible for extraction per the Noun Phrase Accessibility Hierarchy (SU > DO > IO > oblique > GEN > OCOMP), a universal tendency observed across languages. Pragmatic subtypes incorporate topic prominence, where S/A often serve as topics in information structure, though this aligns with the core S/A grouping without altering the basic pattern. Unlike ergative-absolutive alignment, which unites S and P against A, nominative-accusative emphasizes agentivity and subjecthood in transitive and intransitive clauses.[3]Ergative-Absolutive Alignment
In ergative-absolutive alignment, the single argument of an intransitive verb (S) and the patient of a transitive verb (P) share the same morphological marking, known as the absolutive, which typically serves as the unmarked baseline case, while the agent of a transitive verb (A) receives distinct marking, termed the ergative.[3] This pattern manifests in both morphological and syntactic domains; for instance, in the morphology of noun phrases, languages like Dyirbal employ case suffixes where the absolutive is zero-marked for S and P, and the ergative suffix -ŋgu marks A, as in bala ŋayu yaraŋgu ("the boomerang hit me," with ŋayu as absolutive P and yaraŋgu as ergative A). Syntactically, this alignment appears in verbal agreement systems, such as in Sumerian, where hamtu (perfective) verb forms use prefixes to cross-reference the ergative A and suffixes for the absolutive S or P, as in lugal-e é mu-un-dù ("the king built the house," with ergative prefix mu- for A and absolutive suffix -∅ for P).[15] Ergative-absolutive alignment is distributed across approximately 17% of sampled languages, with notable concentrations in certain regions and families. It is common in Australian languages like Warlpiri, Mayan languages such as Yucatec Maya, Tibeto-Burman languages including Tibetan, and isolates like Basque, as well as in parts of the Caucasus, New Guinea, and Austronesian languages; it is rarer in Africa and most of Europe.[3] This geographical spread highlights ergativity's areal and genetic diversity, though it often co-occurs with split systems influenced by tense-aspect or animacy. Subtypes of ergative-absolutive alignment include morphological ergativity, where the pattern is evident in noun case marking and verbal agreement but not in core syntactic pivots, and syntactic ergativity, where S and P function as the privileged arguments for processes like relativization or control. In morphological ergativity, Inuktitut exemplifies this through absolutive case (zero-marked) for S and P on nouns, with ergative -up for A, and verb agreement that cross-references absolutive arguments via suffixes, as in transitive verbs agreeing with both A (via prefix) and P (via suffix). Syntactic ergativity, by contrast, treats S/P as the pivot in clause combining; in Tibetan, for example, the ergative marks A optionally based on volition and valency, but syntactic operations like gapping or relativization pivot on the absolutive S or P, as in constructions where the transitive patient parallels the intransitive subject.[16] Semantic restrictions often condition this alignment, particularly in perfective tenses, where ergative marking applies strictly to animate or agentive As, as seen in Sumerian's hamtu forms versus accusative maru forms.[15] A representative example appears in Basque, an ergative-absolutive language where absolutive is unmarked and ergative ends in -k: gizonak liburua irakurri du ("the man has read the book," with gizonak as ergative A and liburua as absolutive P), paralleling the intransitive liburua erori da ("the book has fallen," with liburua as absolutive S).[17] This contrasts with nominative-accusative systems by unifying S and P in core grammatical behavior.[3]Tripartite and Active-Stative Alignment
Tripartite alignment represents a rare morphosyntactic pattern in which the single argument of an intransitive verb (S), the agent of a transitive verb (A), and the patient of a transitive verb (P) are each marked distinctly through case, agreement, or other grammatical means.[3] This maximal differentiation contrasts with more common alignments that group at least two arguments together, such as nominative-accusative or ergative-absolutive systems. Tripartite alignment occurs in only about 2% of sampled languages worldwide, making it one of the least frequent types, though it appears in isolated pockets across language families.[3] Languages exhibiting tripartite alignment often show this pattern in nominal case marking or verbal agreement. For instance, in Nez Perce (a Sahaptian language of the Pacific Northwest), the S receives nominative marking, the A is marked ergative with the suffix -wá, and the P takes accusative marking with -ním. In Udi (a Northeast Caucasian language), case marking similarly distinguishes all three core arguments: S in the absolutive, A in the ergative, and P in the dative, reflecting a historical development from more layered systems in the East Caucasian family.[18] Such patterns are particularly noted in some Caucasian languages, where complex case inventories allow for fine-grained argument encoding, though full tripartite systems remain exceptional even within the region.[19] Active-stative alignment, also known as split-S or semantic alignment, involves a division of the S argument based on its semantic role: agent-like (SA) arguments of active intransitive verbs pattern with A (often marked as agentive), while patient-like (SP) arguments of stative or non-volitional intransitive verbs pattern with P (marked as patientive).[10] This role-based splitting is motivated by factors such as volitionality, control, or affectedness, leading to verb-dependent marking rather than a uniform treatment of S.[20] Active-stative systems are more common than tripartite ones, appearing in approximately 7% of sampled languages for verbal person marking, with higher concentrations in certain areas like the Americas and Austronesia.[12][10] Representative examples illustrate the semantic sensitivity of active-stative alignment. In Guaraní (a Tupian language of South America), the active intransitive "I run" uses the agentive prefix a- on the verb (a-gwa), aligning SA with A, whereas the stative "I sleep" uses the patientive prefix che- (che-gwa), aligning SP with P.[20] Similarly, in Lakota (a Siouan language of North America), active verbs trigger agreement with SA using subject prefixes (e.g., wa- for first person in "I see"), while stative verbs use object-like affixes for SP (e.g., ma- in "I am seen/tired"). In Austronesia, Acehnese employs a fluid-S variant where S marking can shift based on volitionality: agentive S takes nominative case like A (e.g., "I hit" with active voice), but patientive S takes accusative like P (e.g., "I hurt" with stative voice).[21] These systems highlight how semantic properties like agency drive morphosyntactic choices, often extending to affect verb agreement and case assignment.[22] Active-stative alignment can sometimes overlap with ergative subtypes, where high agentivity reinforces the split, but it fundamentally prioritizes semantic roles over syntactic uniformity.[10]Complex Alignment Systems
Split-S Alignment
Split-S alignment is a type of morphosyntactic system in which the single argument (S) of an intransitive verb is inconsistently aligned with either the agent (A) or patient (P) of transitive verbs, depending on grammatical factors such as tense, aspect, or verb class, rather than semantic role.[23] This results in two subclasses of intransitive verbs: one where S patterns with A (SA) and another where S patterns with P (SP), creating a deviation from pure nominative-accusative or ergative-absolutive patterns.[24] A classic core mechanism involves aspect-conditioned splits, as seen in languages where perfective aspects treat S like A (with ergative marking on transitive A), while imperfective aspects treat S like P (with absolutive marking).[25] Such systems are widespread across diverse language families, including Indo-Iranian languages like those in the Persian branch, Austronesian languages such as Tagalog and Acehnese, and Papuan languages in the Timor-Alor-Pantar family.[23][26] Splits are often conditioned by tense or aspect, with ergative patterns emerging in past or perfective contexts, though they can also depend on verb semantics or nominal hierarchies.[7] In a survey of 39 languages in the Indonesian area (including Austronesian and non-Austronesian/Papuan), 16 exhibited split-S systems, predominantly in eastern Indonesia.[26] Split-S systems can be categorized into subtypes based on the conditioning factors. Hierarchical splits are person-based, following a nominal hierarchy where higher-ranked arguments (e.g., 1st or 2nd person pronouns) pattern S with A, while lower-ranked ones (e.g., 3rd person nouns) pattern S with P; this is common in languages with prominence scales.[23][27] Transitivity-based splits, by contrast, depend on the inherent transitivity or lexical properties of the verb, with "high-transitivity" or agentive intransitives aligning S with A and "low-transitivity" or patientive ones aligning S with P.[24] Prominent examples include Indo-Iranian languages, where past tenses trigger ergative splits: in Hindi, perfective transitive clauses mark A ergatively and P absolutive, with S of agentive intransitives (SA) also ergative and S of patientive intransitives (SP) absolutive, originating from reanalysis of participles.[25] Similarly, in Persian and other Iranian languages, past tenses show a shift to ergative alignment, with compound verbs reanalyzed to produce split-S patterns (e.g., Balochi perfective past indexing agents ergatively).[28] In Austronesian Tagalog, split-S emerges in the voice system, where statives and potentives create a three-way distinction, but binary splits align S with A or P based on verb class and aspect.[29] Papuan Western Pantar exhibits split-S in verbal agreement, where prefixes index S like P for some verbs and like A for others, conditioned by verb semantics.[30] Australian languages often feature split-S tied to noun classes, as in systems where nominal hierarchies (e.g., human vs. non-human classes) determine whether S aligns with A or P, per Silverstein's prominence scales.[23][27] Unlike fluid-S systems, where speakers can choose alignment based on context, split-S alignments are fixed by grammatical rules.[24]Fluid-S Alignment
Fluid-S alignment represents a subtype of semantic alignment in which the single argument (S) of an intransitive verb can be marked either as agentive (aligning with the transitive agent A) or patientive (aligning with the transitive patient P), with the choice determined by the speaker on a per-instance basis rather than fixed lexical classes.[31] This flexibility allows the same intransitive verb to vary its S marking depending on contextual semantics, distinguishing fluid-S systems from more rigid split-S patterns where verb classes dictate marking.[10] Such systems are rare worldwide, occurring primarily in certain Native American languages of the California and southwestern regions, including Pomoan and Yuman families, as well as isolated instances in other areas like the Caucasus (e.g., Tsova-Tush).[10] Unlike grammatically conditioned splits, fluid-S marking emphasizes pragmatic and semantic nuance, appearing sporadically in Australian languages but more prominently in North American contexts where it contrasts with consistent accusative or ergative patterns.[31] The primary factors influencing S marking in fluid-S systems are semantic properties of the event, including the degree of volition, control exerted by the S argument, and the extent of affectedness on the participant.[20] For instance, high volition or control prompts agentive marking (S_A), while involuntary or affected states favor patientive marking (S_P); these choices are not strictly grammatical but reflect speaker interpretation of the scenario.[10] A classic example appears in Central Pomo, a Pomoan language of California, where the intransitive verb meaning "hurt myself" can mark the first-person S either as agentive (if implying self-inflicted action under control) or patientive (if portraying passive suffering).[20] Similarly, in Miwok dialects such as Lake Miwok, motion verbs exhibit S flexibility: a verb like "go" might take agentive marking for purposeful travel but patientive for uncontrolled movement, underscoring how fluid-S systems prioritize semantic roles like agency over syntactic uniformity.[31] This variability highlights implications for semantic role assignment, allowing languages to encode nuanced participant involvement in events. Fluid-S shares semantic motivations with active-stative alignment but offers greater instance-level flexibility.[10]Marking Patterns
Dependent Marking
Dependent marking is a strategy in morphosyntactic alignment where grammatical relations between arguments and predicates are indicated through morphological markers, such as case affixes or adpositions, on the dependent elements—typically nouns or noun phrases—rather than on the head (the verb).[32] This approach contrasts with head marking, where relations are expressed via agreement on the verb. Dependent marking is prevalent across language families and serves to encode core arguments like the single argument of intransitive verbs (S), the agent-like argument of transitive verbs (A), and the patient-like argument (P).[32] In nominative-accusative alignment, dependent marking typically leaves S and A unmarked (in the nominative case), while P receives overt marking, such as the accusative case. For instance, in German, the direct object "den Hund" (the dog, accusative) is marked to distinguish it from the subject, as in "Der Mann sieht den Hund" (The man sees the dog). In ergative-absolutive alignment, the pattern inverts: S and P are unmarked (absolutive), while A is distinctly marked (ergative). Australian languages like Dyirbal exemplify this, where the transitive agent takes an ergative suffix (-ŋgu), as in "bayi yaraŋgu ŋayu banagaŋu" (the man-ERG 1SG.ABS hit-PST), with "yaraŋgu" (man-ERG) marked as A and the S/P forms unmarked.[33] A notable example in passive constructions appears in Latin, where the agent (A) of a transitive verb is marked with the ablative case, as in "Puella a servo liberata est" (The girl was freed by the slave), signaling the agent's role without verbal agreement.[34] Many Australian languages feature extensive noun case systems for dependent marking, often with 10 or more cases to encode not only core arguments but also spatial and semantic relations, reinforcing ergative patterns.[33] This marking strategy facilitates flexible word order, as case indicators on dependents clarify syntactic roles independently of linear position, a feature common in languages ranging from isolating types with adpositions (e.g., English prepositions) to fusional languages with rich inflectional paradigms.Head Marking
Head marking is a morphosyntactic strategy in which the head of a syntactic phrase—typically the verb in clauses—bears affixes or other morphological markers that cross-reference the grammatical roles of its dependent arguments, such as subjects and objects. This contrasts with dependent marking by locating relational information directly on the governing element rather than the modified ones. In head-marking languages, verbs often incorporate pronominal elements for the agent (A) and patient (P) arguments of transitive clauses, as well as the single argument (S) of intransitive clauses, allowing for compact expression of core relations.[35][32] Head-marking patterns can align with various morphosyntactic alignments, including nominative-accusative or ergative-absolutive systems, depending on which arguments the verb agrees with. For instance, in accusative patterns, the verb may agree with S and A but not P, while in ergative patterns, agreement targets S and P (absolutive) but treats A separately (ergative). A classic example of ergative head marking occurs in Mayan languages, where verbs use Set A affixes for ergative (transitive A or possessor) and Set B affixes for absolutive (intransitive S or transitive P). In Kaqchikel (a Mayan language), the perfective form "x-e-in-tz’ët rje’" glosses as 'I saw them', with "e-" (Set A, 1sg ergative) marking the A argument and "in-" (Set B, 1sg absolutive) incorporated for the implied P; intransitive S uses Set B as in "x-Ø-u’l" ('he arrived'). This head-marked ergativity is the default in perfective aspects across Mayan, though non-perfective contexts may show accusative splits.[36][36] Polysynthetic languages like Mohawk (Iroquoian) exemplify head marking through extensive verbal affixes that pronominally encode A and P, often rendering full noun phrases optional. In Mohawk, transitive verbs prefix elements for both arguments; for example, "wa’-ak-nihste’ ra onkwe" means 'I see the man', where "wa’- " is the 1sg prefix for A (subject), "ak- " marks the verb root 'see', and the P is cross-referenced if pronominal or omitted if third-person indefinite, with the full NP "ra onkwe" ('the man') appositional. Intransitive verbs similarly affix for S, as in "wa’-k-en-’" ('I am sleeping'), with "wa’- " (1sg S) and patient-like marking. Salishan languages, such as Halkomelem, extend head marking via noun incorporation, where nominal elements suffix directly to the verb to form complex predicates. For instance, "ní ć e n k’wés-ć e s" translates to 'I burned my hand', incorporating the suffix "-ć e s" ('hand') into the verb stem, treating the possessed noun as a verbal modifier without separate case on dependents.[37][37][38] In Algonquian languages, head marking includes inverse constructions that signal alignment shifts based on a person hierarchy (e.g., 1st > 2nd > 3rd > obviative), where the verb affixes adjust to indicate whether the A outranks the P (direct) or vice versa (inverse). For example, in Plains Cree, a direct form like "ni-ski-si-n" means 'I shake him' (1sg A on 3sg P), but the inverse "ki-ski-s-o-n" means 'he shakes me' (3sg A on 2sg P), with the suffix "-o-" marking the inverse to realign the higher-ranked P as the logical object. This system maintains head marking on the verb while flexibly handling hierarchical prominence. Head marking facilitates compact clauses by integrating argument information into the verb, reducing reliance on word order or adpositions, and is particularly prevalent in the Americas (e.g., 47 consistently head-marking languages in WALS samples) and Papua New Guinea, where it dominates clause-level typology.[39][39][35][32]Locus of Alignment Marking
In morphosyntactic alignment, the locus of marking determines whether grammatical relations between arguments and predicates are expressed on the dependent elements (typically noun phrases), the head (usually the verb), or both. Dependent marking involves affixes or adpositions on nouns to indicate roles, such as nominative for subjects and accusative for objects in many Indo-European languages. Head marking, by contrast, relies on the verb cross-referencing arguments through agreement in person, number, or gender, common in polysynthetic languages of the Americas. Double marking combines these strategies, with nouns bearing case markers and the verb showing agreement; Turkish exemplifies this, as subjects receive nominative case while the verb agrees with them in person and number.[40][41] Mixed systems introduce further complexity by treating specific arguments uniformly across alignment categories, often based on hierarchies like animacy or affectedness. In primary object alignment, prevalent in Bantu languages such as Chichewa, the agent (A) and primary object (P1, typically the beneficiary or most salient patient) receive similar treatment—both can trigger verb agreement and passivize as subjects—while the secondary object (P2) is marked differently, usually without agreement. Dative subject constructions disrupt standard loci by assigning dative case to subjects of non-agentive predicates (e.g., experiencers in verbs of perception or existence), as in Icelandic where "honum leiðast þetta" ('he is bored by this') marks the subject with dative, shifting alignment from nominative-accusative to a quirky case pattern.[42] Typological surveys reveal that dependent marking dominates in clause-level alignment, occurring in 63 of 234 sampled languages (about 27%), followed by head marking (71 languages, 30%), double marking (58 languages, 25%), and zero marking (42 languages, 18%), with zero-marking languages relying on word order or context. At the whole-language level, inconsistent or mixed loci prevail (51% of 236 languages), but dependent marking remains a core strategy in Eurasian and African families, while head marking clusters in the Americas and New Guinea. A key correlation exists with basic word order: SOV structures, the most common globally, strongly favor dependent marking, as case affixes on pre-verbal nouns clarify roles in verb-final clauses, a pattern reinforced in families like Indo-European and Altaic.[40][35][43] Illustrative examples highlight these loci in practice. The Papuan language Yimas demonstrates head-marked ergativity, with no case on nouns but the verb prefixing agreement for the ergative agent (A) and suffixing for the absolutive (S or P), as in "narm-n-tu-k" ('man-ERG-3SG.DU.ABS-1SG.see'), encoding transitive relations solely on the head. In Indo-European languages, the evolution from Proto-Indo-European—which featured both rich dependent marking through inflectional cases and head marking via verbal agreement—to modern Romance languages has involved simplification of case systems, with analytic prepositions increasingly replacing inflectional cases while verbal agreement is retained in reduced form.[44]Theoretical Frameworks
Dixon's Classification
R.M.W. Dixon developed a foundational framework for classifying morphosyntactic alignment systems, initially outlined in his 1979 paper on ergativity and comprehensively detailed in his 1994 book. This typology distinguishes three primary alignment patterns—accusative, ergative, and split—using the universal syntactic-semantic primitives S (subject of an intransitive verb), A (agent-like subject of a transitive verb), and O (patient-like object of a transitive verb). Central to Dixon's model is the role of S as a pivot in syntactic constructions such as coordination, relativization, and control, where S aligns either with A (forming an S/A pivot in accusative systems) or with O (forming an S/O pivot in ergative systems). Split systems, in contrast, exhibit variation in alignment across different grammatical contexts, such as tense, noun phrase types, or semantic features.[45] Dixon's key contributions include integrating semantic roles into the analysis through an agentivity scale, which ranks arguments by degrees of control, volition, and affectedness, influencing how S splits into agentive (Sa, aligning with A) and non-agentive (So, aligning with O) subtypes in certain languages. He further proposed the Nominal Hierarchy (1st person > 2nd person > 3rd person pronouns > proper names > human nouns > animate nouns > inanimate nouns), which predicts splits based on the inherent prominence of noun phrases, with higher-ranked elements favoring accusative marking and lower-ranked ones ergative marking. These elements provide a semantic basis for understanding why pure alignment types are rare, emphasizing instead the interplay between syntax and semantics in real languages.[45] The framework has significant applications in explaining ergative splits in Australian languages. For instance, in Dyirbal, nouns and third-person pronouns follow an ergative pattern (S/O absolutive unmarked, A ergative-marked with -ŋgu), while first- and second-person pronouns exhibit accusative alignment (S/A unmarked, O accusative-marked with -na), aligning with the Nominal Hierarchy's prediction that speech-act participants resist ergative marking due to their high prominence. Similarly, Yidiñ displays a split system with optional ergative marking on nouns (-ŋgu) and accusative on proper names (-ŋya), alongside mixed pivots (S/A for coordination, S/O for relativization), illustrating hierarchy effects in discourse contexts. Dixon's model thus predicts and accounts for such patterns, showing how semantic hierarchies drive deviations from pure ergativity in Australian Aboriginal languages.[45] Despite its influence, Dixon's classification has faced critiques for overemphasizing the S-pivot as the primary diagnostic for alignment, which may oversimplify languages with mixed or weak pivot behaviors and conflict with generative approaches that prioritize argument structure over surface pivots. For example, in Dyirbal and Yidiñ, semantic exceptions in split-S systems (e.g., non-isomorphic agentivity in intransitive verbs) highlight limitations in assuming S's primacy, as full syntactic ergativity remains exceedingly rare across languages. These critiques underscore the need for multi-parameter analyses beyond Dixon's S-focused typology, though his work remains seminal for semantic conditioning of alignment.[46][45]| Alignment Type | Pivot Grouping | Morphological Marking Example | Syntactic Behavior |
|---|---|---|---|
| Accusative | S/A vs. O | S and A: nominative (unmarked); O: accusative (marked) | S/A as subject in coordination (e.g., "The man ran and saw the dog") |
| Ergative | S/O vs. A | S and O: absolutive (unmarked); A: ergative (marked) | S/O as pivot in relativization (e.g., Dyirbal: "the man who the dog bit") |
| Split | Varies by context (e.g., hierarchy, tense) | E.g., pronouns accusative, nouns ergative (Dyirbal) | Mixed pivots, e.g., S/A in coordination, S/O in control (Yidiñ) |