• Aucun résultat trouvé

From Core Knowledge Representations to Linguistic Numbers : A Universal Base for Counting

N/A
N/A
Protected

Academic year: 2021

Partager "From Core Knowledge Representations to Linguistic Numbers : A Universal Base for Counting"

Copied!
2
0
0

Texte intégral

(1)

HAL Id: halshs-01495573

https://halshs.archives-ouvertes.fr/halshs-01495573

Submitted on 25 Mar 2017

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

From Core Knowledge Representations to Linguistic Numbers : A Universal Base for Counting

Pierre Pica

To cite this version:

Pierre Pica. From Core Knowledge Representations to Linguistic Numbers : A Universal Base for Counting . 50 years of Linguistics at MIT, Dec 2011, Cambridge Mass, United States. 2011. �halshs- 01495573�

(2)

1.  ANS and OTS in Mundurucu 2.  Linguistic structure of Mundurucu numerals

3.  Chunks and base

For further information:

pierrepica@gmail.com

http://www.pierrepica.com Mundurucu use of numbers exhibit psychophysics properties of both the

ANS (Approximate Number System), Pica et al., 2004) and OTS (the Object-Tracking System), Pica et al., 2008).

How to reconcile the exact number of syllables (OTS) and the Weber’s law effect (ANS) data in Mundurucu?

4.  Conclusions and prospects

Linguistic structure is an important window for the

study of core knowledge systems (CK) and their

relation to vision. Core Knowledge Systems are

viewed as building blocks of the C/I domain. Taking as a

starting point Mundurucu, a language with few numerals

(up to 5), we suggest that the language faculty satisfies

external constraints in a minimal way.

Spoken or written Mundurucu number words can only refer to approximate quantities,

with an uncertainty that increases with number

(Weber’s law), see Figure 1.

Figure 1 from Pica et al., 2004

Mundurucu number words are long, often having as many syllables as the corresponding quantity, see Table 1. Each syllable refers to an

individual, as perceived in the OTS system.

Figure 2, Give-a-Number task (after Wynn, 1990)

« Give me three seeds! » Stimuli: pûg ,xep xep, ebapûg,

ebadipdip, pûg pôgbi, etc (from Izard et al., 2008)

Age < 3 ½ «Subset Knower»

Can produce sets from numbers up to 1,2 or 3.

Word Reference Syllables

Pûg Between 1 and 2 One

Xepxep Between 1 and 3 Two Ebapûg Between 2 and 7 Three Ebadipdip Between 3 and 8 Four

Table 1

Number words in Mundurucu can be recursively obtained through the application of the operator “other”. “Other” can apply to a constituent

“one object” to yield a structure “(an) other object”.

For example, xepxep can be analyzed as [xep [other xep]]

(with the meaning ‘one, other one’ ).

The operator other is expressed through

syllabic reduplication. The reduplication will be interpreted as a functional category. It follows

that reduplication can be applied only to

“meaningless” syllables.

The structure of number words follows from the above principle and analysis;

e.g. * pûgpûg . At a general level, these facts follow from the constraints on other and its antecedent. The analysis extends to

* xepxep pûg (with the meaning 2+1) if we assume that reduplication cannot take place

inside an additive structure. Ebapûg is completely natural.

It is possible to analyze a numeral such as ebadipdip as a chunk ‘eba’ (lit ‘your two

arms’) followed by another chunk ‘dipdip’, where the first ‘dip’ is interpreted as ‘one object’ and the second ‘dip’ as ‘(an) other one object’.

Crucially, chunks of more than two individuals are not observed. Moreover, any chunk is made of

individuals of the same kind of objects (Feigenson, 2008). This constraint on homogeneity applies to additive structures as well.

It follows from these observations that counting in Mundurucu is «object-specific». It is now possible to analyze ‘pôgbi ’ (lit hand ) in ‘ pûg pôgb i’ (lit one hand(ful),

or 5) not as a chunk , but as an approximate base.

See also ‘xe pxep pôgbi ’ (lit two hands or 10).

Figure 3, xepxep pôgbi

The analysis developed resolves a tension: on the one hand, the number of syllables for numerals until 4 (but crucially not for 5) seems to express the exact cardinality of the numeral; on the other hand, all numerals are interpreted as approximate quantities. The fact that every tracked

individual is expressed is indeed due to external constraints (OTS). These constraints, in turn, explain the approximate meaning of the numerals

(ANS). Linguistic structure satisfy properties of both systems.

Interestingly, examples such as ‘the three of us’, ‘the three of us’, ‘all three of us’, versus * ‘the seventeen of us’, * ‘all seventeen of us’, etc.

might hint that these constraints are universal.

Références

Documents relatifs

We saw that OAL provides different functionalities to assist the linguist during the whole process of lexicon enrichment: a relevant corpus may be crawled from

To support the optimal digitizing system selection according to a given application,the two databasesare coupled with a decision process based on the minimization

Unit´e de recherche INRIA Rennes, Irisa, Campus universitaire de Beaulieu, 35042 RENNES Cedex Unit´e de recherche INRIA Rhˆone-Alpes, 655, avenue de l’Europe, 38330 MONTBONNOT ST

The upper (brown) dots represent the result of Gaudin [22] and coincides with the energy of Lieb’s type-II excitation with zero velocity (momentum π¯ hn) in the model with

LiLa addresses this challenge by building a Linked Data Knowledge Base of linguistic resources (e.g., corpora, lexica, ontologies, dictionaries, thesauri) and NLP tools

Next, there is a color distribution analysis of each layer, which results in a corrected table (matrix) of the color. The next stage of the resulting new three-layer matrix is

virus and virus–host interactions in their cellular context, a human–human protein interaction network containing roughly 70 000 non-redundant protein–protein interac- tions and 10

We also show how one can use this abstract formulation both for giving different and sim- pler proofs for all the known results obtained for the eigenvalues of a power of the