Pre-final draft to appear in:
Cambridge Encyclopedia of the Language Sciences
Patrick Hogan, editor
Cambridge University Press
expected year of publication: 2008

Universals of Semantics

Leonard Talmy
Department of Linguistics and Center for Cognitive Science
University at Buffalo, State University of New York

A semantic universal is any aspect of meaning that is somehow represented in all languages. To first characterize the two terms of this subject separately, semantics pertains to conceptual material as it is organized by language. It ranges from single linguistically represented concepts to principles of conceptual organization.

Universality has properties varying along three parameters. The "level" of a universal is "absolute" (Greenberg’s (e.g., 1963) term) if it is realized in every individual language. But its level is here termed "abstractive" if the universal is part of the language faculty but not of every language. This second abstractive level can in turn be realized in the three following ways, each exemplified below. An abstractive universal is "inventorial" if each language draws its own selection of elements of a certain type from a relatively closed inventory of such elements. An abstractive universal is "typological" if languages fall into a typology on the basis of manifesting one or another out of a relatively small set of (combinations of) certain elements in an otherwise common system. An abstractive universal is "implicational" (Greenberg’s term) if one linguistic feature can appear in a language only if another feature -- itself abstractive -- also appears there.

Another parameter of universality is its "weighting". At the extremes of weighting, a universal is either positive or negative. At such extremes, an absolute universal feature is either present in or absent from all languages, while an abstractive universal feature is either part or not part of the language faculty. But while some features pertaining to an abstractive universal may appear in all or no languages, other features can range from appearing in most down to few languages -- behavior that must also be regarded as part of the language faculty. Traditionally, such non-extremes have been termed "tendencies". But to promote the idea of a gradient hierarchy from positive down to negative, and to highlight the fact that such tendencies are themselves properties of the language faculty, such non-extremes will here be termed "positively tending abstractive universals" or "negatively tending abstractive universals" -- with whatever further modifiers it might take to indicate the degree.

A final parameter of universality is its "subject" -- the particular linguistic phenomenon that manifests the universal property. For a semantic universal, this can range from a specific concept, through a conceptual category or set of such categories, to a conceptual system, and can extend as well to principles of conceptual organization.

All the forms of universality in this framework presumably have a cognitive basis, and this will be addressed below where feasible.

The framework can be illustrated by much within the present author’s work. A positive absolute universal is that the morphemes of every language are divided into two subsystems, the open-class, or lexical, and the closed-class, or grammatical (see Talmy (2000a ch. 1). Open classes have many members and can readily add more. They commonly include (the roots of) nouns, verbs, and adjectives. Closed classes have relatively few members and are difficult to augment. They include bound forms -- inflections, derivations, and clitics -- and such free forms as prepositions, conjunctions, and determiners. Closed-classes can also be implicit, as with word order patterns, lexical categories, and grammatical relations.

A semantic difference correlates with this formal difference. The meanings that open-class forms can express are almost unrestricted, whereas those of closed-class forms are highly constrained. This constraint first applies to the conceptual categories they can refer to. For example, many languages around the world have closed-class forms in construction with a noun that indicate the number of the noun’s referent, but no language has closed-class forms indicating its color. A positive abstractive universal accordingly is that the grammatical morphemes of a language can represent an approximately closed set of conceptual categories, such as those for number, gender, tense, aspect, causality, and status. Excluded are color, and indefinitely many more such as food and religion -- the corresponding negative absolute universal.

A further semantic constraint on grammatical forms is that, even within represented conceptual categories, only certain member concepts can be grammatically expressed and not others. Thus, a positive abstractive universal is that, within the category of number, a bound closed-class form can represent such concepts as `singular’, `dual’, `plural’, and `paucal’, while a free closed-class form can also represent such concepts as `no’, `some’, `many’, and `all’. But the corresponding negative absolute universal is that no language’s closed-class forms represent concepts such as `even’, `odd’, `dozen’, `countable’, or any other concept pertaining to number.

Open-class morphemes are not subject to these same semantic constraints on categories or member concepts. This is shown by the existence of such morphemes as food, even and odd.

However, even open-class morphemes exhibit a few semantic constraints. Thus, on the one hand, the meaning of a morphemic verb can incorporate particulars of aspect that can in turn interact with external closed-class aspectual forms. On the other hand, it is a negative absolute universal that the meaning of no morphemic verb incorporates a tense that can in turn interact with external closed-class tense forms. If not for this exclusion, we might expect to find a verb like (to) went that could be used in a construction like I am wenting to mean `I was going’, or in a construction like I will went to mean `I will have gone’.

Comparably, proper nouns like Manhattan or Shakespeare exist that refer to a specific bounded portion of the space-time continuum. But a negative absolute universal is that there are no "proper verbs" or "proper adjectives" with the same property. That is, both can be "type specific" but they are "token-neutral". Thus, there is never a verb like (to) Deluge referring uniquely to the so-conceived spatio-temporally bounded event of the biblical flood, as in some sentence like: After it Deluged, Noah landed the ark.

Due to the semantic constraints on the closed-class subsystem, the total set of conceptual categories and their respective sets of member concepts that can ever be represented by closed-class forms constitutes an approximately closed inventory. This inventory is universally available. No language has closed-class forms representing all of the conceptual categories and member concepts in the inventory. Rather, each language draws in a unique pattern from the inventory for its particular set of grammatically expressed meanings. Accordingly, this inventory is a positive abstractive universal of the language faculty, not an absolute universal overtly manifested in all languages. In turn, the pattern in which individual languages select their particular sets of grammatically expressed conceptual categories and member concepts from the inventory is governed by a principle of semantic representativeness. No language draws all of its grammatically expressed concepts from one category, say, from aspect alone, but rather draws them from across the range of available categories. The specifics of this principle are not yet clear, but its realization in all languages makes it a positive absolute universal.

Talmy (2006) observes that closed-class forms representing spatial schemas cross-linguistically draw their conceptual categories and member concepts from only a portion of the general inventory, hence, from what can be considered a spatially relevant sub-inventory. For example, out of all the member concepts within the "number" category earlier cited as available to languages for various grammatical specifications, only four ever play a role in closed-class spatial schemas. These are `one’, `two’, `several’, and `many’. Accordingly, the spatially relevant sub-inventory is also a positive abstractive universal, and is embedded within the general inventory, itself having the same type of universality.

Although the main closed-class inventory as a whole is abstractive, that is, with its components merely available for inclusion in individual languages, some components of it might well be represented grammatically in all languages, hence may constitute a positive absolute universal. One candidate might be the concept `negative’ -- along with the conceptual category of polarity to which it belongs. Between positive absolute universals like these and such negative absolute universals as the exclusion of color from grammar, the components of the inventory lie along a gradient hierarchy. Thus, the category of number may be a positively tending abstractive universal, represented in many but perhaps not all languages. And the category of rate with the member concepts `fast’ and `slow’ is a negatively tending abstractive universal, represented in only a few languages.

The next issue is what determines the conceptual categories and member concepts included in the inventory as against those excluded from it. No single global principle is evident, but several semantic constraints with broad scope have been found. One of these, the "topology principle", applies to the meanings -- or "schemas" -- of closed-class forms referring to space, time or certain other domains. This principle excludes Euclidean properties such as absolutes of distance, size, shape, or angle from such schemas, and thus constitutes a negative absolute universal. Instead, these schemas exhibit such topological properties as "magnitude neutrality" "shape neutrality", and "bulk neutrality" -- the corresponding positive absolute universal.. To illustrate magnitude neutrality, the spatial schema of the English preposition across prototypically represents motion along a path from one edge of a bounded plane perpendicularly to its opposite. But this schema is abstracted away from magnitude. Hence, the preposition can be used equally well in The ant crawled across my palm, and in The bus drove across the country. Apparently, no language has two different closed-class forms whose meanings differ only with respect to magnitude for this or any other spatial schema.

Another semantic constraint on concepts available in the inventory pertains to the meanings of conjunctions that head a subordinate clause in a complex sentence (Talmy 2000a ch. 5). Where such conjunctions relate two clauses whose events are in temporal sequence and often also in a cause-effect sequence, with one exception the conjunctions are lexicalized to take the earlier (and causal) event in the subordinate clause, leaving the later (and caused) event in the main clause, and never the other way around. The exception is that -- in addition to an after-type conjunction, which obeys the constraint -- languages can also have a before-type conjunction. Thus, beside We left after we ate, English has We ate before we left. But all other cases follow a negative absolute universal. Thus, beside a because-type conjunction that obeys the constraint, as in English We stayed home because they arrived, English has no inverse conjunction lexicalized to express the hyphenated phrase in *They arrived to-the-occasioning-of-the-event-that we stayed home -- and seemingly no other language does either. Comparably, beside an although-type conjunction, as in We went out even though they arrived, there is never a conjunction lexicalized to represent the hyphenated phrase in *They arrived in-ineffective-counteracting-of-the-event-that we went out. Talmy (2000a ch. 5) bases a cognitive account for this unidirectional lexicalization on an earlier event’s natural function as Ground for a later event as Figure. And it details a similar unidirectionality in conjunctions for the temporal inclusion of one event in another, the contingency of one event on another, and the substitution of one event for another.

Based on their formal and semantic differences, treated so far, a further major finding is that the two types of form classes exhibit a functional difference. In the conceptual complex evoked by any portion of discourse, the open-class forms contribute most of the content, while the closed-class forms determine most of the structure. This division of labor in cognitive function amounts to a positive absolute universal: the open-class subsystem as a whole represents conceptual content and the closed-class subsystem as a whole represents conceptual structure (for illustration, see Talmy 2000a ch. 1). Although individual open- and closed-class forms in general or in a particular sentence may perform the opposite functions, the subsystems overall are universally dedicated to their respective content and structure functions. Further, the function of the closed-class subsystem to structure conceptual content presumably accounts for the negative absolute universals that constrain its semantics. The crucial conclusion -- again a positive absolute universal -- is that the closed-class subsystem is perhaps the most fundamental conceptual structuring system of language.

The main focus so far has been on semantic abstractive universality that involves an inventory, but we now switch to a type that involves a typology. Talmy (2000b ch. 1, 2, 3) examines such typologies for a "extended Motion event" (the capitalized form covers both motion and location). This larger event consists of a Motion event proper and a Co-event that usually represents the Manner or the Cause of the Motion. In turn, the event of Motion consists of four components -- the moving or stationary "Figure", its state of "Motion" (moving or being located), its "Path" (path or site), and the "Ground" that serves as its reference point. What may be a positive absolute universal is that every language has coordinated lexicalization patterns and syntactic constructions to represent all the components of an extended Motion event directly and colloquially over at most two clauses. But the particular lexical and syntactic categories in which the semantic components are characteristically represented is not an absolute universal. Rather, each language has a characteristic pattern of such semantic-formal associations and -- unlike the inventory case where each language is unique -- the patterns range over a relatively small set and so constitute a typology. The "framing" typology is based on where the Path characteristically appears. It divides languages into two main types. In a "satellite-framed" language such as English, the Path is characteristically expressed by a satellite and/or preposition -- like the into in The bottle floated into the cave. But in a "verb-framed" language such as Spanish, it characteristically appears in the main verb -- like the entr´ in La botella entr´ (flotando) a la cueva: `The bottle entered (floating) to the cave’.

Finally, we turn to the implicational type of abstractive universality. It in turn has two forms. In the only form Greenberg posited -- what might be called "other-directed" -- the presence of one feature in a language licenses the presence of a certain different feature. It was seen above that, among subordinating conjunctions that relate two events in time, a `before’-type conjunction was unusual in placing the later event in the subordinate clause. A possible implicational universal is that only if a language has a conjunction expressing `after’ can it have one expressing `before’.

An abstractive implicational universal is "self-directed" if a certain feature that is not an absolute universal but occurs in only some languages always exhibits a certain characteristic -- one from a range of potential characteristics -- when it does occur. This can be seen in the phenomenon of "fictivity", where the meanings of morphemes in a sentence belong literally to one semantic category, but function systematically to represent the opposite semantic category (Talmy 2000a, ch. 2). In the type called "fictive motion", a sentence with morphemes referring literally to motion instead depicts a static scene. In one type of fictive motion, "emanation", the motion formulation evokes the conceptualization of something intangible emerging from a source traveling in a straight line through space, and impinging on a distal object, where in fact nothing can be perceived as moving. Many languages, including English, can use a fictive formulation to represent various types of emanation. For example, English can express a "readiation path", as in Light shone from the sun into the cave; a "shadow path", as in The pole threw its shadow against the wall, or in The pole’s shadow fell on the wall; and a "sensory path", as in I looked into / past the valley.

Not all languages, though, can represent such situations in terms of fictive motion, so that fictivity here is not an absolute universal. Such languages instead tend to use non-fictive formulations like "The sun illuminated the inside of the cave"; "The pole’s shadow is on the wall"; and "I regarded the interior of the valley". But if a language does use fictive motion to depict an emanation, the path of the emanation is always in the same single direction. Thus, radiation is never represented as moving from an object toward the sun, as in *Light shone from my hand onto the sun; nor as moving outward from a third point, as in *The light shone onto my hand and the sun from a point between us. Comparably, a shadow is never represented as moving from the silhouette to the object, as in *The shadow jumped from the wall onto the pole. And a sensory path is never represented as moving from a perceived object to a perceiver acting agentively (though it can be for an unintentional perceiver), as in *That distant valley looked into my eyes. This semantic pattern -- the unidirectionality of fictive emanation -- is thus a positive abstractive universal that is implicationally self-directed.

Further, an "active-determinative" principle appears to govern this universal direction of emanation. Of the two objects, the more active or determinative one is conceptualized as the source. Thus, relative to my hand, the sun is brighter, hence, more active, and must be treated as the source of radiative emanation. My agency in looking is more active than the inanimate perceived object, so I am treated as the source of sensory emanation. And the pole is more determinative -- I can move the pole and the shadow will also move, but I cannot perform the opposite operation of moving the shadow and getting the pole to move -- so the pole is treated as the source of shadow emanation. In turn, this principle might itself derive from the unidirectionality of agency, as detailed further in Talmy (2000a, ch. 2).

Many further semantic universals can be cited. In fact, most semantic findings of cognitive linguistics in general and of Talmy (2000a, b) in particular are universal in character. For example, universals for the representation of force interaction and causality are set forth in Talmy (2000a ch. 4, 7, 8) of temporal aspect in Talmy (2000a ch. 3); and of Figure-Ground organization in Talmy (2000a ch. 5). The domains treated above were chosen for this short article because together they illustrate all the parameters initially outlined for universality.

To mention one further tradition of universalist semantics, the "natural semantic metalanguage" of Wierzbicka (e.g.. 1996) and Goddard (e.g.. 2001) is prominent and extensively developed. This NSM theory posits that a specific set of fundamental concepts -- "semantic primes" -- exists; that it is represented in every language by specific morphemes of that language; and that all other morphemically expressed concepts in the language can be represented by syntactically well-formed combinations of the morphemic primes. In terms of the initial framework, NSM as a whole thus represents positive absolute universality.

References

Comrie, Bernard. 1981. Language universals and linguistic typology. Chicago: University of Chicago Press.

Goddard, Cliff. 2001. Lexico-semantic universals: A critical overview. Linguistic Typology 5(1):1-65.

Greenberg, Joseph. 1963. Some universals of grammar with particular reference to the order of meaningful elements. In Greenberg, Joseph, ed. Universals of Language. Cambridge, MA: MIT Press.

Greenberg, Joseph H. (ed.). 1978. Universals of human language, Vols. 1-4. Stanford, California: Stanford University Press.

Talmy, Leonard. 2000a. Toward a Cognitive Semantics, volume I: Concept structuring systems. i-viii, 1-565. Cambridge: MIT Press.

Talmy, Leonard. 2000b. Toward a Cognitive Semantics, volume II: Typology and process in concept structuring. i-viii, 1-495. Cambridge: MIT Press.

Talmy, Leonard. 2006. The fundamental system of spatial schemas in language. In: From perception to meaning: Image Schemas in Cognitive Linguistics ed. by Beate Hamp. Mouton de Gruyter. pp. 199-234.

Wierzbicka, Anna. 1996. Semantics: Primes and Universals. Oxford University Press.

Zaefferer, Dietmar (ed.). 1991. Semantic Universals and Universal Semantics, volume 12 of Groningen-Amsterdam Studies in Semantics (GRASS). Berlin: Foris.