Semantic Universals

Emmon Bach*
SOAS / UMass (Amherst)
SOAS 11 May, 2010
© Emmon Bach
email: firstinitiallastname at linguist dot umass dot edu

"The best argument in favor of the universality of natural language expressive power is the possibility of translation. The best argument against universality is the impossibility of translation."


  1. Background
  2. Two views of scientific inquiry
  3. From Syntax to Semantics
  4. Time and Space
  5. Kinds, Individuals, Stuff
  6. Pronouns and Deixis
  7. Conclusions
  1. Background
  2. The controversies surrounding Daniel Everett's characterization of the Amazonian language Pirahã and the Evans and Levinson paper about "the myth of language universals" (2009) are just two recent manifestations of a debate about linguistic theory and methodology that is anything but new.

    Everett's article in Current Anthropology (Everett, 2005) and attendant and subsequent commentary were widely publicized, not just in professional journals but in such popular journals as The New Yorker and Der Spiegel and in many newspapers. [Remark: Schadenfreude factor: Chomsky refuted!] The Evans and Levinson paper, with its commentaries and reactions did not receive quite the public press that Everett's paper occasioned, but still was widely discussed and beblogged. We will focus here on the claims and counterclaims about semantics that arise in this context. First, a bit about the methodological background.

  3. Two views of scientific inquiry.
  4. (Bach, 1965; Bach and Chao, 2009, To appear; Croft, 2003; Cinque, 2002).

    1. Baconian / taxonomic
    2. Bloomfield: "The only useful generalizations about language are inductive generalizations." (1933: p. 20)

      Martin Joos [for the record]: "Trubetzkoy phonology tried to explain everything from from articulatory acoustics and a minimum set of phonological laws taken as essentially valid for all languages alike, flatly contradicting the American (Boas) tradition that languages could differ from each other without limit and in unpredictable ways, and offering too much of a phonological explanation where a sober taxonomy would serve as well.

      Children want explanations, and there is a child in each of us; descriptivism makes a virtue out of not pampering that child." (Joos, 1957:96). [EB: my italics, often misattributed in the literature, Joos is characterizing a tradition!]

    3. Keplerian / hypothetico-deductive
    4. A universal is just a general term that takes its meaning from its place in a theory. [Of course, following this route does not preclude claiming universals of the first sort, but refuting or supporting them is tricky!]

      Linguistics differs from chemistry in that each description (grammar) of a language is a theory about that language, while general linguistics is a theory about such theories: Universal Grammar! (in the sense of Montague). [I will avoid this term as it has too much emotional and intellectual baggage at the moment. More in discussion period?]

      Why the second way is better: Compare:

      (i)The planets of the solar system can differ from each other without limit and in unpredictable ways.

      (ii)The planets of the solar system are all made of green cheese.

      The second statement (ii) is a "better" hypothesis. But note that we know it is false!

      Remark: the general cultural literacy about linguistics and language has hardly chamged in 100 years or so.

      Mitchell, 1975[1968] p. 14: Like most inflected languages, OE distinguishes number, case, and gender, in nouns, pronouns, and adjectives.

      Remark: cf. case vs CASE.

      Remark: ideology, temperament, history, sociology of linguistic research.

    5. The effability hypothesis
    6. We take the thesis of effability as a general guide or heuristic which puts conditions of adequacy on the general framework within which languages can vary:

      Effability: Languages are by and large equivalent in their expressive power. (Cf. von Fintel and Matthewson, 2008, Bach and Chao, To appear.)

      To give some content to this claim we need to spell out what "by and large" is supposed to allow.

      Another slogan: All languages are created equal! Some history.

    7. Preliminary example: the NP Universal of Barwise and Cooper (1981)
    8. NP-Quantifier universal: Every natural language has syntactic constituents (called noun-phrases) whose semantic function is to express generalized quantifiers over the domain of discourse. [NP here corresponds to what is now usually called DP, following Abney 1987.]
      Generality universal: Every natural language can express general statements and particular statements.

      Note that this a purely semantic claim, while the NP-Quantifier universal is a claim about syntax AND semantics.

      Note also that it is a substantive claim, as it has been denied -- rightly or wrongly -- by Everett.

      1. Fish swim.
      2. Every fish swims.
      3. Fish usually swim.
      4. Dutchmen make good sailors.
      5. [Port Royal logic: The Art of Thinking]
      6. Fish were swimming.
      7. Compare:
      8. Murgatroyd swims.
      9. Murgatroyd is swimming.

      Possible semantic universals here? Relation between species and individuals, individuals and stages? (Cf. Carlson 1977 and much subsequent literature. More below.)

      Remark: Evans and Levinson's discussion of the NP-Quantifier universal.

  5. From Syntax to Semantics
  6. Semanticists routinely make a distinction between "structural" semantics and lexical semantics. The latter having to do with the meanings of individual words or lexemes. We will be concerned here with both kinds of semantics. It is not clear that the distinction can be cleanly obeyed. For example, the interpretation of examples like those just given (plain vs progressive forms) depends heavily on the semantics of the particular lexical items.

    So: candidates for universals, system of syntactic categories and mappings from categories to semantic types.

  7. Time and Space
    1. A Space Flight
    2. How do we compute the truth value of this sentence:

      1. If Mary had left this morning on her space-flight for Mars, she would now be eating breakfast.
      Necessary ingredients for understanding the meaning of Examples (1):

        Individuals: Mary, Mars, her space-flight,...
      1. Happenings: leaving, eating breakfast,...
      2. Kinds of things: space-flights,...
      3. Relations among things: her = she + Relation (spaceflight), for (Mars),...
      4. Truth-values: true, false, maybe?
      5. Ways that things are or might be: here, now, what might have been,...
      6. Time?
      7. Contexts: value of here, now, Present,...

    3. The structure of time in one model structure
    4. A model structure is a system of possible denotations or semantic values to be used in (one part of) an explication of natural language semantics. Here is one:

      1. BOOL = truth values: {True, False}
      2. A: set of individuals
      3. S: set of worlds
      4. J: set of times
      5. F: set of all functions built from above elements


      • "Henry VIII" is interpreted as a member of A
      • "the prime minister" can be interpreted as a function from situations and times to individuals, e.g. its value at Now is GB?
      • "member of (British) Parliament" is interpreted as a function from individuals to truth-values (equivalently: a set of individuals)
      • dual interpretations: like Frege's distinction between sense and reference
      • cf. prime minister above
      • natural languages as "pragmatic" languages: two stage interpretation from context to denotation to evaluation: theory of "I," "here," "now"

      The set J of times in regimented by a an ordering with what properties? In Montague's PTQ (Montague 1973) the set of times is connected, transitive, and antisymmetric: i.e. any two times are ordered by , and if ti ≤ tj and tj ≤ ti then ti = tj.

      Sentences are evaluated relative to a pair of a world and a time, that is, the set of times is independent of worlds, a questionable assumption.

    5. die ewige Wiederkehr Nietzsche and modern cosmology
      1. Whatever is happening now has happened before and will happen again.

      A localistic alternative: the elements of the model include events, and these events may be (locally) ordered by the relations of precedence and overlap, dropping the connectedness assumption. So Whorf was wrong about SAE assumptions about time, at least in the purely semantic (non-cultural) sense. This allows a more complicated undertanding of Ex. (8) and does not rule out eternal recurrence on semantic grounds.

    6. "grue, bleeexck"etc. Are there linguistic constraints on lexical meanings?
    7. I side here with Humpty Dumpty, at least on lexical meaning and at least some of "structural" semantics. (Explication.)

      "When I use a word," Humpty Dumpty said,..., "It means just what I choose it to mean -- neither more nor less."
      grue: an object is grue if it is green before 2000 and blue thereafter. (Goodman 1955)

      supergrue: an object is supergrue if it is green right now, but blue ever after.

      bleeexck is an adjective which means a combination of qualities that are not expressible in your favourite theory about lexical concepts (e.g. Wierzbicka's Universal Semantics, Jerry Fodor's language of thought, Bishop Wilkins' universal language).

  8. Kinds, Individuals, Stuff.
  9. The general model structure set out above does not make any distinctions in the set of individuals. It is clear that natural languages make lots of finer distinctions within this general realm. As mentioned above, Carlson (1977) initiated (or carried on) a fruitful discussion of various sorts within the general realm and there has been a lot of research in the last decades investigating the semantic (and syntactic) properties of various distinctions here. One big division that seems pervasive is the distinction between (ordinary) individuals and events (eventualities) and their subkinds: states, processes, events. Another is that between Kinds and Individuals, Individuals and Stages. Another area that has been the subject of a lot of research is that centering around mass and count terms. There is rich material here for investigating the semantic and syntactic properties of such distinctions, both over and covert (in the sense of Whorf (1945). Here is a small selection of examples illustrating some of these domains:

    1. Katrina occurred in 2005. (a huricane)
    2. Harry is (*being) in New York.
    3. Jill is running in the London Marathon (right now / this year).
    4. Garcia hits a home run.
    5. The Siberian tiger is endangered.
    6. Simba is endangered.
    7. I hate raccoons because they are destroying my garden.
    8. There was mud on the floor.
    9. There are three muds (kinds of mud) in Muswell Hill.
    10. El Tigre is on the prowl tonight.

    Excursus: about oolichans (= candlefish Thaleichthys pacificus). "The common names of this fish have a somewhat confusing relationship. The name `candlefish' derives from the fact that it is so fat during spawning, with up to 15% of total body weight in fat, that if caught, dried, and strung on a wick, it can be burned as a candle. This is the name most often used by early explorers. The name "eulachon" (occ. "oolichan", "oulachon", "uthlecan", etc.) is from the Chinookan language and the Chinook jargon based on that language. The name "hooligan" appears to have been derived from "eulachon" by similarity with the English slang term for a ruffian or scoundrel which gained currency in the late 19th century." (Wikipedia s.v. eulachon)

    1. It didn't come again this year.
    2. Why "it"? Cf. rice, wheat, pease vs peas, etc

  10. Pronouns and Deixis
  11. Functions of pronouns (pronominal affixes, etc).

    1. deixis
      1. She is my friend.
      2. Oowek'ala (Northern Wakashan) Hilda Smith†: `his/her/their pencil: (cf Bach, 2006, 2007)

        item-dem (inv) inst owner-dem (inv)
        -ga (here by me) -k (here by me)
        ITEM -ax̌ (there by you) (-c) -s -x̌ʷ (there by you) (-c)
        -a (yonder) -i (yonder)
        -xdi just gone -ki just gone

      3. k̓adayug(ʷ)asx̌ʷ -ga -s -x̌ʷ
      4. `the visible pencil right here belonging to that visible person right there by you'
      5. k̓adayug(ʷ)ack -ga -c -s -k
      6. `the hidden pencil right here belonging to this visible person right here by me'
      7. k̓adayug(ʷ)ackc -ga -c -s -k -c
      8. `the hidden pencil right here belonging to this invisible person right here by me'
      9. k̓adayug(ʷ)asx̌ʷ -ga -s -x̌ʷ
      10. `the visible pencil right here belonging to that visible person right there by you'

        BUT: grammatical meaning vs lexical meaning:

      11. Who is the man drinking the Martini? It's Harry but it's a ginger ale.
      12. A semantic universal!

    2. anaphora
      1. John said he would follow us.

    3. quantification: with and without bound variables
      1. Every Dutchman believes he is a good sailor.
      2. Compare:
      3. Dutchmen believe they are good sailors.

      Quantification without bound variables (Jacobson 1999, Steedman 2000).

      Roots of quantification: expressions of totality (Sapir 1930, Bach 1995).


      1. y̓ay̓uḡʷáulh `rain all the time' y̓úgʷa `rain'
      2. t̕lat̕laḡʷáulh `red blanket' (red all over) t̕láqʷa `copper, red'
      3. tlalagáulh `paddle all the way' tláka `to paddle'
      4. ʼAgámi láʼexc̓i. `they (remote, visible) all left.'
      5. ʼagam- -i `all + 3-remote-vis-subj'

      Relations between linguistic categories and semantic values many-many?

  12. Conclusions: Whorf revisited.
  13. Long ago, when I first read Whorf's influential essay, "An American Indian view of the universe" (Whorf, 1956), I found myself thinking: What he says about Hopi may or may not be true, but I am skeptical about what he says about English and other Standard Average European (SAE) languages and their metaphysics. Whorf expresses the view that languages like English embody a basically Newtonian view of time and space, while Hopi is congenial to a much more relativistic conception of the universe. But Einstein was no Hopi.

    Quite a few years later but still quite a few years ago I ventured into such questions myself from the point of view of model-theoretic semantics (Bach, 1981, 1986). Model-theoretic semantics requires setting up some basic apparatus for the structure of denotations: individuals, truth-values, functions of various sorts, possible worlds and times. Doing this requires making some choices about the basic stuff of the world that is used in the models.

    With Ken Hale: (1975: 296 f.)

    I think that the correct way to understand the Australian counting systems is as follows: conventionalized counting systems, i.e., numerals, are for the most part lacking, but counting itself is not lacking, in the sense that the principle of addition which underlies the activity of exact enumeration is everywhere present. In fact, I would like to argue that counting, in this sense, is unversal, and whether or not a conventionalized inventory of numerals exists in a given language depends upon the extent to which exact enumeration is of practical use or necessity to the people who speak the language.[2] One might look upon the Walbiri lack of conventionalized numerals as a gap in the inventory of cultural items -- since the principle which underlies counting is present. filling the gap is a rather trivial matter. This view is entirely compatible with the observation that the English counting system is almost instantaneously mastered by Walbiris who enter into situations where the use of money is important (quite independently of formal Western-style education, incidentally).

    What I am suggesting here is that certain cultural iatems can be said to be universal even though they may not be included in the inventoary of cultural items for particular communities. This is not a contradiction if one bears in mind that what is universal is the concept, not some conventionalized manifestation of it.

    Hale goes on to apply this way to thinking to syntax, in particular to the presence and nature of relative clauses (adjoined and embedded), a topic that is highly relevant to Everett's discussion of embedding and recursion. Recommended reading!

    A Final General Observation (cf. Bach, 1974)

    Variation within languages tends to approach variation across languages

    1. Classifiers: six head of cattle, twelve pieces of furniture,...
    2. Ideophones: He ran lickety-split down the hill,...

    This is why sometimes the best way to study a language is to study other languages!

Footnote: *A good part of this presentation represents joint work with Wynn Chao. All problems are due to me or gremlins.


Abney, Steve. 1987. The English Noun Phrase in its Sentential Aspect. Ph. D. dissertation MIT. Bach, Emmon. 1965. Structural Linguistics and the Philosophy of Science. Diogenes 51, 111-128. (Also appeared in French.) Bach, Emmon. 1974. Syntactic Theory. New York: Holt, Rinehart and Winston, Bach, Emmon. 1986. Natural language metaphysics. In R. Barcan Marcus, G.J.W. Dorn and P. Weingartner, eds. Logic, Methodology, and Philosophy of Science VII (Amsterdam et al.: North Holland, 1986), pp. 573-595. Bach, Emmon. 1995. A note on quantification and blankets in Haisla. In Emmon Bach, Eloise Jelinek, Angelika Kratzer, and Barbara H. Partee, eds., Quantification in Natural Languages (Dordrecht: Kluwer). Bach, Emmon. 2006. Paradigm regained: deixis in Northern Wakashan. SOAS Working Papers in Linguistics 14: 267-281. Bach, Emmon. 2007. Deixis in Northern Wakashan. In Peter Austin and Andrew Simpson, eds., Endangered Languages [Linguistische Berichte Sonderhefte 14], pp. 253-265. Bach, Emmon and Wynn Chao. 2009. Semantic universals and typology. In Chris Collins, Morten Christiansen and Shimon Edelman, eds., Language Universals (Oxford: Oxford University Press). Pp. 152-173. Bach, Emmon, and Wynn Chao. To appear. Language universals from a semantic perspective. To appear in Claudia Maienborn, Klaus von Heusinger, and Paul Portner, eds. Semantics: An International Handbook of Natural Language Meaning. Berlin: Mouton de Gruyter. Barwise, Jon and Robin Cooper. 1981. Generalized quantifiers and natural languages. Linguistics and Philosophy 4: 159-219. Carlson, Greg N. 1977. Reference to kinds in English. Ph.D. dissertation: University of Massachusetts, Amherst. Cinque, Guglielmo. 1999. Adverbs and Functional Heads: A Cross-Linguistic Perspective. New York / Oxford: Oxford University Press. Cinque, Guglielmo, ed. 2002. Functional Structure in the DP and the IP: the Cartography of Syntactic Structures, Vol I. Oxford / New York: Oxford University Press. Croft, William. 2003. Typology and Universals. Second edition. Cambridge: Cambridge University Press. Evans, Nicholas and Stephen C. Levinson. 2009. The myth of language universals: Language diversity and its importance for cognitive science. Behavioral and Brain Sciences 32, 429-492. (with comments by a number of cognitive scientists and linguists) Everett, Daniel L. 2005. Cultural constraints on grammar and cognition in Pirahã. Current Anthropology Volume 46, Number 4 (August-September). Pp. 621-646. (inclusive of comments and reply) Comments by Brent Berlin, Marco Antonio Gonçalves, Paul Kay, Stephen Levinson, Andrew Pawley, Alexandre Surrallé, Michael Tomasello, Anna Wierzbicka. Goodman, Nelson. 1955. Fact, Fiction, and Forecast. Cambridge, MA: Harvard UP, 1955. [2nd ed. Indianapolis: Bobbs-Merrill, 1965. 3rd. ed. Indianapolis: Bobbs-Merrill, 1973. 4th ed. Cambridge, MA: Harvard UP, 1983.] Hale, Ken. 1975. Gaps in grammar and culture. In M. D. Kinkade, K. L. Hale and O. Werner, eds., Linguistics and Anthropology, in Honor of C. F. Voegelin (Lisse: Peter de Ridder Press), pp. 295-315. [SOAS Library: A417/358274] Jacobson, Pauline. 1999. Towards a variable-free semantics. Linguistics and Philosophy 22:117-184. Joos, Martin, ed. 1957. Readings in Linguistics. Chicago and London: University of Chicago Press. Mitchell, Bruce. 1975. A Guide to Old English. Oxford: Basil Blackwell. [Reprinted with corrections: 2nd edition: 1968] Montague, Richard. 1973. The proper treatment of quantification in ordinary English. (PTQ) In Richard Montague, Formal Philosophy, edited by Richmond Thomason (New Haven: Yale University Press), pp. 247-270. Sapir, Edward. 1930. Totality. Linguistic Society of America, Language Monographs, No. 6. 28 pp. Steedman, Mark. 2000. The Syntactic Process. Cambridge, Massachusetts / London: MIT Press. von Fintel, Kai, and Lisa Matthewson. 2008. Universals in Semantics. The Linguistic Review 25 (2008), 139-201 Available on Whorf, Benjamin Lee. 1945. Grammatical categories. Language 21:1-11. Reprinted in Whorf, 1956, pp. 87-111. Whorf, Benjamin Lee. [1956] An American Indian model of the universe. In Whorf, 1956, pp. 57-64. Whorf, Benjamin Lee. 1956. Language, Thought, and Reality: Selected Writings of Benjamin Lee Whorf. Edited by John B. Carroll (Cambridge, Mass.: MIT Press). Wierzbicka, Anna. 2005. Comment on Everett, 2005. Current Anthropology, 46: 641.