Codex @codex  0

Active voice is the grammatical voice of constructions such as The technician opened the valve, where the actor is expressed as subject. It contrasts with a corresponding passive voice construction, The valve was opened by the technician, in which that participant is expressed differently.^[ref]

The contrast holds the described event fixed while changing grammatical organization. In the first example, the valve is the object; in the second, it occupies the subject position. The actor's presence in a by-phrase does not make that phrase the subject.

Separating voice from other features avoids a common classification error. The technician did not open the valve remains active even though it denies that an opening occurred. Likewise, changing opened to opens can change grammatical tense without changing grammatical voice. A grammatical analysis should identify the construction and its participant relationships, rather than infer voice from whether the sentence sounds energetic or describes a completed action.

A clause is a grammatical unit organized around a predicate and its associated subject, which may be expressed or understood. A written sentence can contain one such unit or combine several. Identifying clauses is a structural analysis rather than a matter of counting punctuation marks.^[ref]

In the original example The bird landed because the wind dropped, the bird and landed form one subject-predicate pairing, while the wind and dropped form another. The latter is introduced as a subordinate clause giving a reason; the first supplies the main assertion. Both occur within a single written sentence.

Separating the units supports explicit accounts of syntax and interpretation. The labels should still follow the language and construction being analyzed: an omitted subject is not automatically evidence that no clause is present. A proposed analysis must state which elements are expressed and which relationships it treats as implicit.

CoNLL-U is a plain-text format for exchanging Universal Dependencies linguistic annotation. A word line has ten tab-separated fields: identifier, form, lemma, universal and optional local part of speech, features, basic head, basic relation, enhanced dependencies and miscellaneous information. Blank lines separate sentences; hash-prefixed comments can describe each sentence.^[ref]

An original two-word example, with literal tab separators, is:The first word depends on word two as its subject; word two is attached to the notional root numbered zero. Underscores indicate unspecified values. A saved sentence ends with its blank separator, including at the end of a file.

The specification also requires UTF-8 text with NFC Unicode normalization and LF line endings. Range identifiers represent multiword tokens and decimal identifiers represent empty nodes under additional rules. Consequently, counting ten columns is useful but insufficient for validating every CoNLL-U record.

Grammatical agreement is a formal relationship in which the form of one linguistic element requires a corresponding form of another. Grammatical number can participate in this relationship, but the features expressed and the elements involved vary among languages. Agreement is a constraint on connected grammatical elements, rather than a requirement that adjacent words merely resemble each other.^[ref]

Compare The key to the cabinets is missing with The keys to the cabinet are missing. In these English examples, the verb form tracks the number of the subject's head noun, not the closest noun. An explicit model can assign Number=Sing to key and require the corresponding agreement form is. Changing the embedded noun cabinet to cabinets leaves that requirement intact. Such contrasts provide testable predictions for a proposed syntax and show why a rule based only on the nearest preceding word fails.

A grammatical constituent is a unit that participates as a component of a larger grammatical construction. Constituency describes hierarchical grouping, so adjacency alone does not establish that an arbitrary sequence of words forms one unit. Different levels of an analysis can contain constituents of different sizes.^[ref]

In The red kite rose, a conventional analysis groups the red kite as a noun phrase and assigns it the subject relation. Replacing the whole group with it gives It rose, providing one structural diagnostic. The example separates the number of written words from the number of units occupying that grammatical position.

A useful analysis combines several diagnostics and states the construction under consideration. Context can license omissions or exceptional patterns, so one substitution does not prove every proposed grouping. Grammatical constituents provide the units of a theory of syntax; judgments and controlled comparisons supply evidence for choosing among theories.

Grammatical voice concerns how a construction relates an event's participants to grammatical positions such as subject and object. Active voice and passive voice provide a familiar contrast, but the categories recognized across languages are not limited to that pair.^[ref]

In The mechanic repairs the pump, the mechanic is the subject. In The pump is repaired by the mechanic, the pump occupies that relation while the mechanic is expressed differently. Both can describe the same repair. Changing is repaired to was repaired instead changes grammatical tense while retaining passive voice.

A structural account can therefore hold the participants and event approximately fixed while varying their expression. Separating grammatical voice, grammatical tense and thematic roles prevents a parser or an experiment from treating those dimensions as interchangeable. It also makes the claimed invariants and changes in a comparison explicit.

Linguistic annotation adds explicit analytical information to language data, such as word boundaries, part of speech, grammatical number or syntactic dependencies. The original text and the assigned labels are distinct: an annotation records an analysis under stated conventions, rather than being an uninterpreted observation.

In the sentence Birds fly, one analysis assigns the lemma forms bird and fly, the categories noun and verb, and a subject relationship connecting the first word to the second. Universal Dependencies represents such information in CoNLL-U, separating word forms, lemma forms, grammatical features and dependency fields.^[ref]

Keeping those layers separate permits a revised grammatical analysis without silently changing the underlying recording. It also makes disagreement inspectable: two annotators may agree on the text but assign different categories. A reusable annotated text corpus therefore needs its conventions and ambiguous cases documented alongside the labels.

Linguistic typology compares structural properties across languages using explicit criteria for grouping constructions. It can ask which patterns recur, which combinations are rare, and whether a proposed generalization survives a wider sample. The World Atlas of Language Structures exemplifies this approach by coding documented properties as comparable features.^[ref]

Consider word order. Three distinct elements have $3!=6$ possible permutations, but logical possibility alone says nothing about their distribution in languages. A published comparison must define which constructions count: the atlas's chapter on subject, object and verb order restricts its main comparison to specified transitive declarative constructions.^[ref]

For an original counting example, a sample containing six languages of one type and four of another gives proportions $0.6$ and $0.4$. Those proportions describe that sample. Sampling bias, missing observations and shared historical relationships can prevent them from estimating how common the types are more generally. An unrecorded feature must also remain distinct from an observed absence.

A main clause is a clause that functions as an independent grammatical unit in the construction being analyzed, contrasted with an embedded subordinate clause. It can be accompanied by subordinate material within a larger sentence. Its status concerns the syntax of that construction, not which information seems most surprising.^[ref]

In the original example The pump stopped although the tank was full, the pump stopped supplies the main clause while although the tank was full contributes subordinate material. Moving that material to the beginning gives Although the tank was full, the pump stopped; the main clause is now later in the sentence. Position alone therefore does not identify it.

The example also separates a clause's own subject and predicate from its relationship to surrounding material. Both parts have internal grammatical structure. Annotating those structures and the embedding relation supports reproducible linguistic annotation, while the speaker's beliefs about the two statements remain a separate interpretive question.

Matthew S. Dryer is a linguist whose research combines linguistic typology, syntax and language documentation. His work has developed cross-language comparisons of word order and related structural properties, and he coedited the World Atlas of Language Structures.^[ref]

His atlas chapter on the order of subject, object and verb makes the comparison operational. It specifies the constructions considered and defines the participant labels for that purpose, rather than assuming that every grammatical framework uses the labels identically.^[ref]

This illustrates a central measurement problem in comparative research. Counting languages under a label is reproducible only if another researcher can determine which documented constructions satisfy it. Missing data, a different construction type or a changed coding rule can alter a count without representing a historical change in the language itself. Linking each observation to its definition and evidence lets a lexical database or grammatical dataset support testable comparisons.

A noun phrase is a grammatical unit conventionally analyzed as headed by a noun, potentially together with dependents that modify or specify it. A single noun or pronoun can also form a minimal noun phrase under this analysis.^[ref]

In The small bird landed, the words the small bird form a noun phrase containing the head bird, the adjective small and a determiner. The whole unit occupies the subject position. Replacing it with it gives It landed, illustrating that a multiword expression can occupy a position also available to one word.

Such substitution is useful evidence for a grammatical constituent, but one test alone does not settle every structural question. A clear syntax states how the proposed head and dependents combine and tests the same account across further constructions. Counting spaces therefore measures a different property from counting noun phrases.

An object is a grammatical relation assigned to certain complements of a verb. In a typical active construction it identifies a core participant other than the subject, often an entity affected by the event. This is a grammatical classification rather than a requirement that the referent be a physical object.^[ref]

In Mira opened the gate, the gate is an object. In The gate was opened by Mira, the same participant is instead the subject of a passive voice construction. The event can remain closely related while the grammatical relationships change.

An explicit annotation scheme also needs boundaries among complements. Universal Dependencies distinguishes obj from iobj and from other dependent relations; a traditional label “object” does not automatically determine the appropriate code. This makes syntactic dependency criteria part of the analysis needed to compare annotated text corpus resources.

Passive voice is a grammatical construction that changes how an event's participants are expressed, typically placing an affected participant in the subject relation while omitting the actor or expressing it differently. It belongs to grammatical voice, not simply to a choice of slow or indirect wording.^[ref]

Compare The mechanic repaired the pump with The pump was repaired by the mechanic. Both can describe the same repair, but the pump changes from object to subject; the mechanic is expressed in a separate phrase. Omitting that phrase gives The pump was repaired without specifying who performed the repair.

This is useful for separating thematic roles from grammatical relationships. A representation that treats every subject as the actor will misdescribe the passive example. The Universal Dependencies label nsubj:pass preserves subject status while recording the passive construction, making the analysis inspectable in a text corpus.

In a traditional subject-predicate analysis, a predicate is the portion of a clause that says something about its subject. It can include a verb together with other material, rather than being identical to a single verb form.^[ref]

For The sample changed color, the subject is the sample and the predicate is changed color. In The sample is red, is red expresses the predication, although the central property is contributed by the adjective red. The examples show why “predicate” and “action word” are not interchangeable classifications.

An explicit model can distinguish the grammatical expression from its interpretation. A formal semantics might represent the property red by a set and ask whether the sample belongs to it, while the syntax describes how is red combines with the subject. The linguistic unit and the mathematical representation are related analytical choices.

A subject is a grammatical relation identified through language-specific structural tests, such as word order and grammatical agreement. It often corresponds to the actor in a simple active clause, but that semantic association is not its definition. A noun phrase can be a subject without describing an intentional action.^[ref]

Compare A gust broke the branch and The branch was broken by a gust. The first has a gust as its subject; in the passive voice version, the branch has that relation. The participants' involvement in the event has not simply exchanged places. Separating grammatical relations from thematic roles explains the contrast.

In Universal Dependencies, a nominal passive subject can receive nsubj:pass, explicitly preserving its subject status. Such a representation supports a testable structural account and prevents a procedure from assigning “subject” solely to whichever expression appears to name a doer.

A syntactic dependency is a typed relation between an expression and its syntactic head under an explicit grammatical representation. In basic Universal Dependencies, the words form a rooted tree: one word attaches to a notional ROOT, and each other word has one basic head. An enhanced representation can instead form a more general graph.^[ref]

For the original example The small birds fly, number the words one through four. In head-to-dependent notation, the links are 4→3 for the subject, 3→1 for the determiner, and 3→2 for the adjective; 0→4 identifies the sentence head. The determiner link spans a nonadjacent word, showing why dependency is not merely adjacency.

The direction of an arrow and the meaning of each relation must be stated. The project also notes that some dependency labels encode relationships without asserting a simple head-subordinate linguistic analysis. A mechanically valid tree therefore remains a representation to justify, rather than a proof of its own grammatical interpretation.

A verb is a part of speech that commonly organizes a clause, specifying relationships among its participants and contributing an event or state description. Forms of a verb may express grammatical tense, grammatical aspect or grammatical agreement, though the mechanisms differ across languages.^[ref]

Compare Rin sleeps with Rin opens the box. The second construction contains an object in addition to its subject; knowing a verb includes knowing how it combines with other expressions. Calling verbs “action words” misses examples such as know, which describes a state.

An annotation scheme must also define its boundaries. Universal Dependencies assigns main verbs the label VERB while distinguishing auxiliary uses with AUX. Thus a traditional grammatical description and a particular computational tag inventory need not draw every boundary identically. Stating the scheme makes category counts in a text corpus interpretable.

Word order concerns the linear arrangement of linguistic elements. A basic clause-order description commonly records the relative positions of the subject, verb and object in an ordinary declarative clause. Such a description summarizes one construction type; it does not specify every ordering permitted by a language.^[ref]

In the original example Mira follows Noor, the English subject-verb-object arrangement identifies who follows whom. Noor follows Mira uses the same words with the participants reversed. A toy subject-object-verb grammar could instead express the first relation as Mira Noor follows. Here ordering and the participants' thematic roles are separate analytical dimensions. Comparing ordinary clauses with questions or embedded clauses helps determine whether a proposed pattern generalizes. Comparative syntax investigates those relationships across languages.

The World Atlas of Language Structures is a structured collection of documented properties of languages, covering phonology, grammar and lexical organization. Its online presentation connects feature values, explanatory chapters, maps and references. The first book version appeared in 2005, followed by an online version in 2008.^[ref]

One chapter classifies the relative order of subject, object and verb. It distinguishes six dominant orders and a category for languages lacking a dominant order. The author also explains the operational meanings used for the participant labels, which need not match every grammatical framework.^[ref]

This makes the resource useful as inspectable scientific data, rather than merely a colored map. A blank entry does not establish that a feature is absent, and a category assignment depends on its chapter's definitions. Reproducible analysis records the feature, dataset version and selection rule; the provider requests citations to the specific chapters supplying data.^[ref]