Meaningful Blurs: the sources of repetition-based plurals in ASL


In several sign languages, plurals can be realized with unpunctuated or punctuated repetitions of a noun, with different semantic implications; similar repetition-based plurals have been described in some homesigns and silent gestures. Unpunctuated repetitions often get approximate ‘at least’ readings while punctuated repetitions typically correspond to ‘exactly’ readings. The prevalence of these mechanisms could be thought to be a case in which Universal Grammar does not just specify the abstract properties of grammatical elements, but also their phonological realization, at least in the visual modality. We explore an alternative in which punctuated and unpunctuated repetitions arise from general properties of iconic representations. On an empirical level, we argue that in ASL (American Sign Language), punctuated and unpunctuated repetitions are unlikely to be an exclusively grammatical mechanism, as they can be found with purely iconic (word-free) representations. On a theoretical level, we argue for a modular account with three components. First, repetition-based plurals can create a simplified pictorial representation. Second, unpunctuated repetitions give rise to pictorial vagueness, resolved by way of quantification over precisifications. Third, a pragmatic process involving strategic reasoning maps these vague representations onto a set of candidate linguistic meanings, including some ‘at least’ plural readings that are best expressed by unpunctuated repetitions.

This is a preview of subscription content, access via your institution.


  1. 1.

    By ‘word-free representations’, we mean improvised gestural representations identified by the native ASL consultant (and co-author) as corresponding to no conventional lexical form. We explore cases of repetition with modification to make it even less likely that a grammatical operation is at work (since to our knowledge standard pluralization does not involve such modifications of a lexical sign).

  2. 2.

    Without the assumption that nouns hold true of atoms and non-atoms alike, one would need to posit an operation of pluralization of the noun before it can hold true of non-atoms.

  3. 3.

    We use the term ‘consultant’ when we discuss information obtained in elicitation sessions (and thus prior to writing the article).

  4. 4.

    As Cremers and Chemla (2017) write, that graded inferential judgments "may help detect otherwise hidden effects". This was borne out for our ASL consultant (who has years of experience with quantitative acceptability judgments): he noted in the past that the inferential part of his task was made easier, not harder, by using quantitative judgments of inferential strength. Without these, he had to reflect at length about how to categorize judgments of intermediate strength; the quantitative method allowed for less arbitrary decisions in such cases.

  5. 5.

    This method should of course be assessed rigorously in the future, but this holds of other elicitation methods as well—and ours has the advantage of transparency (since the raw written data are made available in the Supplementary Materials).

  6. 6.

    For some examples, see the discussion by William Vicars at (retrieved on November 6, 2019).

  7. 7.

    One should not overinterpret the ‘at least four’ inference found in this case and reported in (6). One might expect instead an ‘at least (approximately) three’ reading because there are three iterations. But as one can check by watching the video, the number of iterations is just very difficult to count in this case. In Context 1, there is an ‘at least four’ reading alongside the ‘at least three’ reading, as can also be seen in (6). Still, we don’t know why Context 1 and Context 2 do not give rise to entirely parallel results—nor whether the difference is significant (in view of the limitations of our methods).

  8. 8.

    Note that the inferential possibilities were clearly understood with their pragmatic implications. For instance, in (4)e there was a high endorsement of the ‘exactly three’ reading and a lower endorsement of the ‘approximately 3’ and ‘at least three’ readings, no doubt because these were interpreted to exclude the more informative ‘exactly 3’ situations.

  9. 9.

    There are different possible realizations of ASL G. This one is close to what is represented in (7) (the drawing is from, retrieved on October 17, 2019).

  10. 10.

    As Lamberton notes, the circle represented on the fingers in (11) is slightly bigger than the shape he would use for the small object classifier (for him, the diameter of the circle represented inside the thumb and index finger might be 25–50% smaller in the classifier use). In addition, the small object classifier would be accompanied by a facial expression.

  11. 11.

    Our drawing (from the same source as in fn. 10) is not consistent between and : the former is given from the signer's perspective, the latter from the addressee's perspective (see the pictures in (16) for the ‘real’ addressee perspective).

  12. 12.

    One helpful remark made by the consultant (= [JL 19.10.10]) was phrased as follows: "The signs used might represent abstract shapes rather than actual letters, particularly since the intermediate shapes are clearly not letters. However, the first and last shapes strongly invoke the letters G and L and that rules for me."

  13. 13.

    Since we do not know of other studies of such repetitions with modifications, our discussion leaves several questions open, including (i) how they semantically compare to standard plurals in various environments, and (ii) how they interact with other categories, such as adjectives.

  14. 14.

    The case of numerals in English turns out to be complicated: as discussed in Spector (2013), one of the leading current theories is that numerals have an at least reading (in particular under downward-monotonic environments), but that they associate with exhaustivity operators more freely than other constructions, hence the appearance of exactly readings even in downward-monotonic environments. We revisit this question in relation to ASL repetition-based plurals at the end of Sect. 4.2.

  15. 15.

    The reason is that IF-clauses behave like downward-monotonic environments, which both license Negative Polarity Items such as ever and suspend implicatures, as illustrated in (i):

    (i) a. If you ever go to China, you should visit Shanghai.
      b. If you invite Robin or Sam, I'll give you $20.
    => if you invite Robin and Sam, I'll give you $20.
  16. 16.

    Some care is needed, however, as any semantic differences might be erased if the context somehow facilitates an existential reading. Concretely: suppose one predicate, call it =4, means ‘exactly 4 circles’, and the other, call it ≥4, means ‘at least 4 circles’. If the target sentence is analyzed as in (i), with an existential quantifier within the IF-clause and a distributive predicate, the semantic difference between the two expressions will be erased. Some of the examples we investigated in our fieldwork are arguably like this, and we focus in the text on those that are not.

    (i) a. If ∃x =4(x) & x can be seen, I'll give you $20.
      b. If ∃x ≥4(x) & x can be seen, I'll give you $20.
  17. 17.

    This problem pertains to the contrast between two types of ‘easy to count’ repetitions, slow unseparated and slow separated ones. A further issue for this theory is why unpunctuated repetitions, i.e. fast unseparated repetitions, should fail to give rise to an embedded implicature.

  18. 18.

    Greenberg (2021) is committed to the ‘only if’ part but not necessarily to the ‘if’ part, because further principles might enrich the content of a picture. In case some situations are plausible while others are extremely implausible, the latter might be excluded from consideration when computing the truth conditions of pictures. (Thanks to G. Greenberg, p.c., for discussion of this point.)

  19. 19.

    Greenberg (2014) proposes (in our terms) to define pictorial truth in terms of pictorial truth-of, but we will not be concerned with this reduction in the present piece.

  20. 20.

    This paper was already written when Greenberg (2019) appeared. It provides a semantics for tagging, a hybrid of symbols and pictures, used for instance in maps and in labelled pictures. We leave for future research an investigation of potential connections between the present piece and Greenberg’s analysis of tagging.

  21. 21.

    Here and throughout: pictures are modified from

  22. 22.

    We write ‘almost never’ rather than ‘never’ because there could be special cases to consider. For instance, with marking rule we will adopt in Sect. 9.2, a picture may conceivably be ‘silent’ on what happens in part of the situation of evaluation. If so, a different picture may be true of the same situation relative to the very same viewpoint (e.g. if the second picture is identical to the first one except that it provides information on a component of the situation that the first one is silent on).

  23. 23.

    Technically, the problem is that we ought to ensure that the mereological sum C1+C2+C3 of the three relevant parts of the array can only denote 1 or 2 cups, whereas the sum of the denotations of C1, C2 and C3 includes 3 cups.

  24. 24.

    Still, the analogy with the behavior of a whole unpunctuated repetition in ASL shouldn't be exaggerated. It might be through grammatical means that a plural discourse referent is introduced by the entire set of repetitions. By contrast, iconic means might be needed to introduce singular discourse referents at the edges. This possible claim was tested by Schlenker and Lamberton (2019) with a paradigm that included (i). The idea was that plural reference to the entire set of sentences described by (i) could be effected grammatically, while singular reference to an individual sentence should require an iconic interpretation and an edge. As noted in that earlier paper, "the first sentence evaluated on its own only provides weak iconic information. Adding the second clause with a right-edge possessive pronoun strengthens the iconic inference in (…) [(i)a)], but the effects are not entirely clear.”


    ‘Yesterday you wrote sentences. The meaning of the right-most sentence/of all thesentences was vague.’

    a. 6.3 __ =SENTENCE-rep3horizontal … = POSS-right

    3/3 judgments: the sentences are more likely to be in a row when the 2nd clause is added

    b. 7 __ =SENTENCE-rep3horizontal… = POSS-arc

    1/3 judgment: the sentences are more likely to be in a row when the 2nd clause is added

    (ASL, 34, 2256c,d; 3 judgments)

  25. 25.

    We add to (52)(iii) a maximality condition to make sure that all parts that are available for pointing are taken into account.

  26. 26.

    Note that the mixed marking rule yields a partial redundancy between conditions (53)(i) and (53)(ii). Take for instance a circular punctuated repetition of manual G’s, i.e. , as in (40). By (53)(i), the denoted group d will have to satisfy the condition that [[ ]]c, s[X→d](d) = true, or in other words: d is a (singular or plural) group of G’s. By (53)(ii), the 's in signing space have to pictorially represent G’s, hence the same result. Future research could seek to eliminate the redundancy by doing without condition (53)(i), but one point should be kept in mind: (53)(i) makes reference to a context and an assignment function, whereas (53)(ii) doesn't. Thus if a repeated noun has a context- or assignment-sensitive component, getting rid of (53)(i) might create problems.

  27. 27.

    Special thanks to B. Spector (p.c.) for several helpful discussions, and to an anonymous reviewer for urging that we clarify the initial presentation of the model.

  28. 28.

    The reason we do things in this way is that we take A0 to be entirely naive and to just go by the iconic information she receives. Within a standard RSA model, one might expect the priors in (63) to already play a role at this point, in the following sense: the ‘state of the world’ that the addressee A0 seeks to recover is just the meaning that the speaker has in mind; A0 has two sources of information about the state of the world (i.e. about the meaning the speaker seeks to convey): (i) the priors (= which meanings the speaker is likely to want to express); (ii) the information conveyed by (possibly blurry) repetitions. Here we solely consider (ii) at the level of A0.

  29. 29.

    We will refine the analysis in Sect. 9 to explain why separated slow repetitions give rise to an ambiguity (despite the fact that they are easy to count and should be classified as punctuated).

  30. 30.

    This depends on one's specific view on numeral semantics. The problem we refer to will arise with the rather natural semantics in (i), in which the effect of the numeral 4 is to introduce a maximal group of objects that satisfy the noun and the predicate (this is akin to the ‘reference set’ discussed in connection with anaphora in Nouwen 2003, Chapter 3). If N-iter is made of 3 slow unseparated iterations of TROPHY, it could not be a picture (part) true of a group of exactly 4 trophies, unless some trophies are somehow hidden.

    • (i) [[[4 N-iter]x P]]c, s = true iff for some group d,

    • (a) d is the maximal group d′ such that [[N]]c, s[X→d′](d′) = true and [[P]]c, s[X→d′] = true,

    • (b) d contains exactly four objects, and

    • (c) N-iter is tolerantly true of d.

  31. 31.

    This analytical direction need not conflict with the marking rule in (73)b if the 3 iterations of TROPHY depict a (possibly typical) subpart of the denoted group. In (75)a,b, the hand does not go down between the iterations, and this is presumably responsible for the inference that the trophies are spread out (by (73)b, the areas through which the hand goes without tracing TROPHY are ones in which there is nothing, hence the ‘spread out’ inference). Things are different in (72)b, where the hand goes down between the iterations, with the result that the marking rule in (73)b is silent about what there is between the depicted objects.

  32. 32.

    An ASL sentence cited in part in Davidson (2014) (Figure 3a) appears in (i):

    • (i) THERE CANDLE CL-CY GLOBE CL:5(claw)Z

      ‘There is a candle and a globe.’

  33. 33.

    Davidson (2014) sketches an account based on the interaction between Questions under Discussion and classifiers (which bias the question towards the configuration of the objects). This is congenial to but distinct from the pictorial analysis envisaged here. As in our discussion of (75), however, one issue to keep in mind is that a picture need not depict the entirety of the situation if some objects are hidden from view; this might allow some ASL native signers to accept the underinformative statement.

  34. 34.

    Tangentially, a reviewer asks about further uses of repetitions in ASL. Lamberton can think of these: (i) First, a case that is almost certainly unrelated to the present investigation: a single repetition (i.e. a total of 2 iterations) is a standard and well-known nominalization strategy in ASL, and serves for instance to distinguish CHAIR (nominal, 2 iterations) from SIT (verbal, 1 iteration). (ii) As in English, repetition can serve to intensify, as in I never never never did that. (iii) Far closer to our topic, there are temporal uses of repetitions: to say that someone won many trophies in their career, one might sign TROPHY repeatedly from one's shoulder down to a neutral space—hence: many trophies were won over time. We leave these diverse uses for future research. 

  35. 35.

    Several directions could be explored in future research. First, as suggested by J. Kuhn (p.c.), it would be useful to create stimuli by splicing one and the same realization of an onomatopoeia in several linguistic contexts so as to ensure that the repeated sounds in (80)a,b (or (80)c,d) are genuinely identical. Second, it would be useful to investigate nominal (rather than pluractional) uses of repeated onomatopoeias so as to have a possibly closer point of comparison for ASL plurals. In a different context, Guerrini and Migotti (2019) used repeated onomatopoeias to evoke several champagne bottles (the sound was evocative of a cork popping). We are not sure how to realize these sounds in an unpunctuated fashion, but this direction would be worth exploring.

  36. 36.

    Retrieved on November 23, 2019 from

  37. 37.

    We also said nothing about the potential role of the speed of repetition per se. In Schlenker and Lamberton (2019), it was mentioned that, all other things being equal, greater speed might be associated with larger quantities. In the present theory, speed only plays an indirect role, by making the iterations harder to count. The theory might have to be refined to account for a more direct relation between speed and quantities.

  38. 38.

    In greater detail: Assume w satisfies [3]. Setting E = ∪P'∈π(P) P', we have that for each precisification P' of π(P), P' entails E, hence w ∈ E, i.e. [4]. Now assume [4]. Thus for some P* ∈ π(P), w ∈ P*. It follows that for each proposition E such that for each precisification P' of π(P), P'E, P*E and therefore w ∈ E.

  39. 39.

    Before recording the paradigm, we discussed its acceptability with one of the consultants.


  1. Abner, N., Namboodiripad, S., Spaepen, E. & Goldin-Meadow, S. (2015). Morphology in child homesign: Evidence from number marking. Slides of a talk given at the 2015 Annual Meeting of the Linguistics Society of America, Portland, Oregon.

  2. Bergen, L., Levy, R., Goodman, N. D. (2016). Pragmatic reasoning through semantic inference. Semantics & Pragmatics, 9.

  3. Byrne, R. W., Cartmill, E., Genty, E., Graham, K. E., Hobaiter, C., & Tanner, J. (2017). Great ape gestures: Intentional communication with a rich set of innate signals. Animal Cognition, 20(4), 755–769.

    Article  Google Scholar 

  4. Chierchia, G., Fox, D., & Spector, B. (2012). Scalar implicature as a grammatical phenomenon. In C. Maienborn, P. Portner, & K. von Heusinger (Eds.), Semantics: An international handbook of natural language meaning (Vol. 3, pp. 2297–2331). Berlin: de Gruyter.

    Google Scholar 

  5. Cobreros, P., Egré, P., Ripley, D., & van Rooij, R. (2012). Tolerant, classical, strict. Journal of Philosophical Logic, 41(2), 347–385.

    Article  Google Scholar 

  6. Cohn, N. (2013). The visual language of comics: Introduction to the structure and cognition of sequential images. London: Bloomsbury.

    Google Scholar 

  7. Coppola, M., Spaepen, E., & Goldin-Meadow, S. (2013). Communicating about quantity without a language model: Number devices in homesign grammar. Cognitive Psychology, 67, 1–25.

    Article  Google Scholar 

  8. Cormier, K. (2002). Grammaticization of indexic signs: How American Sign Language expresses numerosity. PhD dissertation, University of Texas, Austin.

  9. Cremers, A., & Chemla, E. (2017). Experiments on the acceptability and possible readings of questions embedded under emotive-factives. Natural Language Semantics, 25(3), 223–261.

    Article  Google Scholar 

  10. Davidson, K. (2014). Scalar implicatures in a signed language. Sign Language & Linguistics, 17(1), 1–19.

    Article  Google Scholar 

  11. Davidson, K. (2015). Quotation, demonstration, and iconicity. Linguistics and Philosophy, 38(6), 477–520.

    Article  Google Scholar 

  12. Davidson, K., & Mayberry, R. I. (2015). Do adults show an effect of delayed first language acquisition when calculating scalar implicatures? Language Acquisition, 22(4), 329–354.

    Article  Google Scholar 

  13. Emmorey, K., & Herzig, M. (2003). Categorical versus gradient properties of classifier constructions in ASL. In K. Emmorey (Ed.), Perspectives on classifier constructions in signed languages (pp. 222–246). Mahwah NJ: Lawrence Erlbaum Associates.

    Chapter  Google Scholar 

  14. Feldstein, E. (2015). The development of grammatical number and space: Reconsidering evidence from child language and homesign through adult gesture. Manuscript, Harvard University.

  15. Fine, K. (1975). Vagueness, truth and logic. Synthese, 30, 265–300.

    Article  Google Scholar 

  16. Fischer, S. (1973). Two processes of reduplication in the American Sign Language. Foundations of Language, 9, 469–480.

    Google Scholar 

  17. Franke, M., & Jäger, G. (2016). Probabilistic pragmatics, or why Bayes' rule is probably important for pragmatics. Zeitschrift für Sprachwissenschaft, 35(1), 3–44.

    Google Scholar 

  18. Goodman, N. D., & Frank, M. C. (2016). Pragmatic language interpretation as probabilistic inference. Trends in Cognitive Sciences, 20(11), 818–829.

    Article  Google Scholar 

  19. Graham, K. E., Hobaiter, C., Ounsley, J., Furuichi, T., & Byrne, R. W. (2018). Bonobo and chimpanzee gestures overlap extensively in meaning. PLoS Biol, 16(2), e2004825.

    Article  Google Scholar 

  20. Greenberg, G. (2013). Beyond resemblance. Philosophical Review, 122, 2.

    Article  Google Scholar 

  21. Greenberg, G. (2014) Reference and predication in pictorial representation. Handout of a talk given at the London Aesthetics Forum, 19 February 2014.

  22. Greenberg, G. (2019). Tagging: Semantics at the iconic/symbolic interface. In Julian J. Schlöder, Dean McHugh & Floris Roelofsen (Eds.), Proceedings of the 22nd Amsterdam Colloquium 2019 (pp. 11–21). Amsterdam: ILLC.

  23. Greenberg, G. (2021). Semantics of pictorial space. Review of Philosophy and Psychology.

  24. Guerrini, J. & Migotti, L. (2019). Musical gestures in the typology of linguistic inferences. Talk given at the workshop on Linguistic Investigations Beyond Language, Berlin. 11 March 2019.

  25. Hyde, D. (1997). From heaps and gaps to heaps of gluts. Mind, 106(424), 641–660.

    Article  Google Scholar 

  26. Klima, E. S. & Bellugi, U. (1979). The signs of language. Cambridge, MA: Harvard University Press.

  27. Kuhn, J. (2015). Cross-categorial singular and plural reference in sign language. PhD dissertation, New York University.

  28. Kuhn, J., & Aristodemo, V. (2017). Pluractionality, iconicity, and scope in French Sign Language. Semantics & Pragmatics, 10(6), 1–49.

    Article  Google Scholar 

  29. Liddell, S. K. (2003). Grammar, gesture and meaning in American Sign Language. Cambridge: Cambridge University Press.

    Book  Google Scholar 

  30. Nash, D. G. (1986). Topics in Warlpiri grammar. New York: Garland.

    Google Scholar 

  31. Newman, P. (2012). Pluractional verbs: An overview. In P. Cabredo Hofherr & B. Laca (Eds.), Verbal plurality and distributivity (pp. 185–210). Berlin: De Gruyter.

    Google Scholar 

  32. Nouwen, R. (2003). Plural pronominal anaphora in context. PhD dissertation, Utrecht University.

  33. Pfau, R., & Steinbach, M. (2006). Pluralization in sign and in speech: A cross-modal typological study. Linguistic Typology, 10, 49–135.

    Article  Google Scholar 

  34. Schlenker, P. (2011). Donkey Anaphora: The view from Sign Language (ASL and LSF). Linguistics and Philosophy, 34(4), 341–395.

    Article  Google Scholar 

  35. Schlenker, P. (2014). Iconic features. Natural Language Semantics, 22(4), 299–356.

    Article  Google Scholar 

  36. Schlenker, P. (2019). What is super semantics? Philosophical Perspectives, 32(1), 365–453.

    Article  Google Scholar 

  37. Schlenker, P. (2020). Gestural Grammar. Natural Language & Linguistic Theory, 38(3), 887–936.

  38. Schlenker, P. (2021). Iconic Presuppositions. Natural Language & Linguistic Theory, 39(1), 215–289.

  39. Schlenker, P., & Lamberton, J. (2019). Iconic plurality. Linguistics and Philosophy, 42, 45–108.

    Article  Google Scholar 

  40. Schlenker, P., Lamberton, J., & Santoro, M. (2013). Iconic variables. Linguistics and Philosophy, 36(2), 91–149.

    Article  Google Scholar 

  41. Spector, B. (2013). Bare numerals and scalar implicatures. Language & Linguistics Compass, 7(5), 273–294.

    Article  Google Scholar 

  42. van Fraassen, B. C. (1966). Singular terms, truth value gaps and free logic. Journal of Philosophy, 63, 481–495.

    Article  Google Scholar 

  43. Varzi, A. (2003). 1995, supervaluationism and paraconsistency. In D. Batens (Ed.), Frontiers in paraconsistent logic (pp. 279–297). Baldock: Research Study Press.

    Google Scholar 

  44. Zucchi, A. (2011). Event descriptions and classifier predicates in sign languages. Presentation at FEAST (Formal and Experimental Advances in Sign language Theory) in Venice, June 21, 2011.

Download references


We are very grateful to three reviewers for Linguistics and Philosophy for extraordinarily constructive and helpful comments, which led to several improvements. Editor Patrick Grosz was also very helpful. Kathryn Davidson and Benjamin Spector made numerous suggestions and objections which helped improve the paper.

P.S.'s acknowledgments For discussion and references, I am indebted to Kathryn Davidson, Paul Egré, Carlo Geraci, Gabe Greenberg, Jeremy Kuhn. Special thanks to Benjamin Spector for particularly illuminating discussions of the pragmatic part, and to the audience of the LINGUAE seminar (Paris, April 23, 2020) for helpful suggestions. For judgments on onomatopoeias in English, I am very grateful to Brian Buccola, Jonah Katz, Jeremy Kuhn, Nathan Klinedinst, Matthew Mandelkern, and Lyn Tieu. I also received much input on comics from Jonathan Bonhomme and Claire Schlenker. Many thanks to Lucie Ravaux for checking all averages and for putting the raw scores in the Excel file made available in the Supplementary Materials, as well as for working on the bibliography.


This research received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (Grant Agreement No 788077, Orisem, PI: Schlenker). Research was conducted at Institut d'Etudes Cognitives, Ecole Normale Supérieure - PSL Research University. Institut d'Etudes Cognitives is supported by grants ANR-10-IDEX-0001-02 and FrontCog ANR-17-EURE-0017.

Author information




Philippe Schlenker initiated this research, constructed all examples in consultation with Jonathan Lamberton, and developed the analysis. Jonathan Lamberton was the ASL consultant for the initial phase of the work. When it was written, he checked or provided transcriptions and translations, and discussed important aspects of the analysis. Any theoretical discussion among co-authors occurred only after the data were collected and the first version of the article was written, and primarily by email because the authors were not on the same continent. The separation between the two phases of the work was intended to minimize the risk of a ‘theoretical contamination’ of sign language judgments.

Corresponding author

Correspondence to Philippe Schlenker.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material


Appendix I: Blurry pictures: truth via entailments vs. bivalent tolerant truth

We reiterate in (83) our definition of bivalent tolerant truth for blurry pictures, and in (84) the definition of truth via entailments for blurry pictures. For readability we boldface propositions, and thus if P is a picture (evaluated relative to a viewpoint v and a system of projection S), we will write as P the set of worlds that make P true.

(83) Bivalent tolerant truth for a blurry picture
Relative to a world w, a viewpoint v and a system of projection S, a picture P associated with precisifications π(P) = {P1, P2, …} is tolerantly true iff for at least one Pi in π(P), projS(w, v) = Pi; tolerantly false otherwise.
(84) Truth via entailments - case of a blurry picture
Relative to a world w, a viewpoint v and a system of projection S, a picture P associated with precisifications π(P) = {P1, P2, …} is true* iff for each proposition E such that P entails E, E is true in w; false* otherwise.

In this case, bivalent tolerant truth yields the same result truth via entailments, as stated in (85).

(85) Let P be a picture evaluated relative to a world w, a viewpoint v and a system of projection S, and let π(P) = {P1, P2, …} be the set of its precisifications.
P is true* in w if and only if P is bivalently tolerantly true in w.


Relative to a viewpoint v and a system of projection S:

P is true* in w iff [1] for each proposition E such that P entails E, w ∈ E
  iff [2] for each proposition E such that for each precisification P′ of π(P), P′ entails E, w ∈ E
  iff [3] for each proposition E such that for each precisification P′ of π(P), P′E, w ∈ E
  iff [4] w ∈ ∪P′ ∈ π(P) P′Footnote

In greater detail: Assume w satisfies [3]. Setting E = ∪P'∈π(P) P', we have that for each precisification P' of π(P), P' entails E, hence w ∈ E, i.e. [4]. Now assume [4]. Thus for some P* ∈ π(P), w ∈ P*. It follows that for each proposition E such that for each precisification P' of π(P), P'E, P*E and therefore w ∈ E.

  iff [5] for some P′ ∈ π(P), w ∈ P′
  iff [6] P is tolerantly true in w.

Appendix II: On Nicaraguan homesigners

Both the iconic and the pragmatic components of our analysis could be expected to apply to other systems, in particular other sign languages, and also homesigns. As a reviewer notes, the production data of Nicaraguan homesigners reported in Coppola et al. (2013) might seem to raise a worry for our analysis. In a production experiment, homesigners “were videotaped describing ten 10- to 20-s vignettes with varying numbers of objects and events”. The authors coded three types of number marking: “Finger Extensions (FEs) were handshape configurations in which each extended finger indicated a member of the target set”; “Punctuated Movements (PMs) were a series of discrete movements, each referring to an entity or action in the vignette”; “Unpunctuated Movements (UMs) were movements produced in rapid succession with no clear break between them”. The latter two categories seem to correspond rather closely to the punctuated and unpunctuated repetitions we described for ASL. In view of the model of Sect. 8, we would expect that, when they are used, punctuated repetitions should only have ‘exactly’ reading. This is roughly consistent with the data in (86): when punctuated movements were used, there were roughly as many iterations as there were objects to describe. What might be concerning, on the other hand, is that the number of unpunctuated iterations does not seem to depend on the number of objects to be described.

(86) Use of finger extensions (FE), punctuated movements (PM) and unpunctuated movements (UM) in a production experiment on Nicaraguan homesigners (Coppola et al. 2013)

There are two issues to be addressed. One is the apparent absence of ‘approximately’ readings of unpunctuated repetitions. Since this is not the dominant reading we derive in (68) (in essence, because we postulated that such a reading would be used less often than the ‘at least’ reading), we might posit that it is either unavailable or not deemed useful in this production situation and/or in this language. With 2-3 unpunctuated iterations, the ‘at least’ readings we derive in (68) are correctly predicted to be true in the situations in which they are used (for present purposes, we treat two unpunctuated iterations along the same model as the three unpunctuated iterations, with an ‘at least 2’ reading). But this leaves open another question: why do subjects not produce ‘at least 3’ readings (with 3 unpunctuated iterations) when there are 3 objects or more, ‘at least 4’ readings (with 4 unpunctuated iterations) when there are 4 objects or more, etc?

There are several possible directions to explore; but an important general lesson is that we would need to take into account cost differences between different unpunctuated repetitions: when n < m, repn (meaning ‘at least n’) is both less informative and less costly than repm (meaning ‘at least m’), which involves more iterations.

  1. (i)

    One possibility to explain the homesign data is that the latter just don't allow for more than 2-3 unpunctuated iterations. But from the perspective of our iconic analysis, this is entirely stipulative.

  2. (ii)

    Another possibility is that the preference for minimal numbers of unpunctuated repetitions is an effect of the task. We could reason as follows: the subjects are asked to describe a scene. If they think that precision matters, they will go for ‘exactly’ readings and thus for punctuated repetitions. If they think that precision doesn't matter, they will go for the least costly (i.e. shortest) unpunctuated repetition compatible with the scene, hence using in effect ‘at least 2-3’ whenever this is true, including for large numbers of objects to describe. On this view, the same results would be expected to arise with ASL-signing subjects faced with the same task.

  3. (iii)

    Yet another possibility is that there is a difference between homesigns and ASL with respect to the cost of the iterations, with more iterations being somehow more costly in homesigns than in ASL. Why this should be is unclear, however.

This question ought to be further investigated in more sophisticated models than the one developed in this piece.

Appendix III: Mini-survey on repeated onomatopoeias

We repeat in (87) the target paradigm involving repeated onomatopoeias, from (80).

(87) Context: the addressee is playing a shooter video game.
a. In the next 10 seconds, you should [prh - prh - prh - prh]slow, pauses.
b. In the next 10 seconds, you should prh-rep4fast, no pauses.
c. If in the next 10 seconds you [prh - prh - prh - prh]slow, pauses, I'll give you $20.
d. If in the next 10 seconds you prh-rep4fast, no pauses, I'll give you $20.

We tested this paradigm by way of audio files and an online survey with 6 consultants (all native speakers of American English, all linguists, including one working on gestures and one working on sign language).Footnote 39 Acceptability and (open or quantitative) inferential questions were modeled after those we used in ASL, with quantitative questions displayed in (88). Average acceptability scores (with 7 = best) and inferential scores (with 7 = strongest) are reported in (89). They should be interpreted with caution because there was a lot of variation among consultants (see the Supplementary Materials).

(88) Questions
a. Acceptability: How acceptable is the sentence you heard on a 7-point scale (with 1 = worst, 7 = best)?
  b. Quantitative inferential questions for (80)a,b
How strongly do you draw the following inferences? (with 1 = no inference; 7 = strongest inference)
In the next 10 seconds, the addressee should shoot…
…exactly 4 times.
…approximately 4 times.
…at least 4 times.
…at least 5 times.
  c. Quantitative inferential questions for (80)c,d
Assuming that the shots are spaced appropriately (if relevant), how strongly do you infer that the speaker is committed to giving the addressee $20 in case the addressee shoots…
…exactly 4 times.
…approximately 4 times.
…at least 4 times.
…at least 5 times.
(89) Inferential results (averages, mini-survey of 6 linguists, see the Supplementary Materials)
Dominant inferences (i.e. those with the highes endorsement for a given sentence) are boldfaced.
Sentences Acceptability Number of objects that the repetitions stand for
=4 ±4 ≥4 ≥5
a. [prh - prh - prh - prh]slow, pauses 5.5 5.7 3.7 4.8 1
b. prh-rep4fast, no pauses 6.5 2 3.3 3.8 2
c. [prh - prh - prh - prh]slow, pauses 5.2 5.7 2.3 4.5 3.2
d. prh-rep4fast, no pauses 6 4.5 4.5 3.8 3.5

In unembedded contexts and under if, the strongest reading of a punctuated repetition (= (89)a,c) is an ‘exactly’ reading. Inferences about unpunctuated repetitions are surprisingly weak, possibly because some consultants obtained a reading on which shooting once with a machine gun gave rise to a burst of shots. Here it is particularly informative to consider the 6 responses to 2 open questions, reproduced in (90)-(91). Most answers mention multiple shots and/or a machine gun (unlike in the case of punctuated repetitions; see the Supplementary Materials for full answers), although not necessarily with an ‘at least 4’ threshold.

(90) About (87)b: What should the addressee do in the next 10 seconds? (provide a description, not necessarily limited to the number of times the addressee should shoot)
1. Shoot rapidly, at least four times
2. fire a machine gun
3. The addressee should fire a burst of shots. I'm imagining an automatic-style gun, where when you pull the trigger (even without holding it for very long), a burst of several shots (around 3-4) get fired very fast.
4. The addressee should shoot.
5. fire a short burst of shots
6. Shoot rapidly as necessary.
(91) About (87)d: Under what conditions must the speaker give the addressee $20? (provide a description, not necessarily limited to the number of times the addressee should shoot)
1. Provided he fire a number of shots
2. fire two pistols alternatively, one in each hand
3. Minimally, the addressee has to fire a burst of around 4 shots. Beyond that, it depends on more context.
4. if the addressee fires their gun / shoots (not necessarily at a target?)
5. fire a short burst of about four shots
6. If the player fires at least one burst of shots, or possibly a single shot.

It is worth noting, however, that Consultant 4 apparently obtained an ‘at least 1’ reading on which the unpunctuated repetition can be satisfied by a single shot. (Consultants also commented on possible effects of intonation, which we did not seek to control for, and differences in realization between one occurrence of the punctuated repetition and the other.)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Schlenker, P., Lamberton, J. Meaningful Blurs: the sources of repetition-based plurals in ASL. Linguist and Philos (2021).

Download citation


  • Sign language semantics
  • Iconicity
  • Plurals
  • Unpunctuated repetitions
  • Punctuated repetitions
  • Pictorial semantics
  • Iconic semantics
  • Super semantics