Berent, Bat-El, Brentari, Dupuis & Vaknin-Nusbaum (2016) @cite{berent-bat-el-brentari-dupuis-vaknin-nusbaum-2016} #
The double identity of linguistic doubling. Proceedings of the National Academy of Sciences 113(48). 13702--13707.
Twelve experiments demonstrating that the interpretation of doubled forms (XX) depends on (i) morphological context and (ii) the speaker's L1 morphological system. The key finding is a 2x2 cross-linguistic dissociation between English and Hebrew speakers:
| English (no productive redup) | Hebrew (redup for diminutives) | |
|---|---|---|
| Plurality | Prefer XX (Exps 2, 4a, 8a) | Weak/no XX pref (Exp 10a) |
| Diminutive | No XX preference (Exp 12) | Prefer XX (Exp 11a) |
Gradient vs. categorical #
The experimental data are gradient: participants show stronger or weaker XX preferences across conditions, measured by rating differences and reaction times. The OT model here gives categorical predictions (reduplication wins or XY wins). The categorical predictions capture the direction of the preference in each cell — which condition shows an XX advantage — not the continuous magnitude of the effect.
Mechanism: positive and negative transfer #
Positive transfer: if the L1 expresses function f morphologically, the speaker can interpret XX as morphological reduplication for f, yielding a doubling preference (XX > XY).
Negative transfer: if the L1 uses reduplication for other functions but specifically NOT for f, the speaker has positive evidence that reduplication != f. This blocks the reduplication interpretation for f, even if f is morphologically marked by other means (e.g., Hebrew marks plurality by suffixation, not reduplication — so Hebrew speakers do not interpret XX as plural reduplication).
Formalization #
The doubling framework (DoublingParse, DoublingGrammar,
realizeMorphAvailable) is defined in Theories/Phonology/Doubling.lean.
This file defines L1-specific DoublingGrammar instances for English and
Hebrew and proves the four cells of the dissociation table as OT theorems.
@cite{berent-bat-el-brentari-dupuis-vaknin-nusbaum-2016}
English morphological knowledge relevant to doubling.
English marks plurality morphologically (dog-s) but does not have productive reduplication for any function, and does not have productive diminutive morphology (booklet, doggy are semi-productive at best).
Equations
- One or more equations did not get rendered due to their size.
Instances For
Hebrew morphological knowledge relevant to doubling.
Hebrew marks both plurality (sefer -> sfarim 'book -> books') and diminutives morphologically. Crucially, Hebrew uses reduplication specifically for diminutives (seleg -> slaglag 'snow -> puppy') but NOT for plurality (which uses suffixation).
Equations
- One or more equations did not get rendered due to their size.
Instances For
English: REALIZE-MORPH is available for plurality. English marks plurality morphologically and has no productive reduplication at all, so there is no negative transfer.
Hebrew: REALIZE-MORPH is available for diminutives. Hebrew uses reduplication specifically for diminutives. Positive transfer: Hebrew speakers interpret XX as diminutive reduplication.
The four cells of the dissociation table follow from the transfer predictions above. When REALIZE-MORPH is available, the morphological ranking applies and reduplication wins. When unavailable, the phonological ranking applies and XY (nonidentical) wins.
English + plurality: reduplication wins (Exps 2, 4a, 8a). English speakers show a gradient XX preference (higher ratings, faster RTs) when signs are paired with homogeneous object sets in a plurality context. The categorical prediction captures the direction of this gradient effect.
English + diminutive: XY wins (Exp 12). English speakers show no XX advantage for diminutive signs because English lacks productive diminutive morphology. The model predicts XY wins categorically; the data show absence of the XX preference seen in the plurality condition.
Hebrew + plurality: XY wins (Exp 10a). Hebrew speakers show weak/no XX advantage for plural signs because Hebrew uses reduplication for diminutives but NOT plurality — negative transfer blocks the reduplication parse. The model predicts XY categorically; the data show attenuation or absence of the XX preference relative to the diminutive condition.
Hebrew + diminutive: reduplication wins (Exp 11a). Hebrew speakers show a gradient XX preference for diminutive signs because Hebrew has productive reduplicative diminutives — positive transfer makes the reduplication parse available. The categorical prediction captures the direction of the effect.
The 2x2 cross-linguistic dissociation: English and Hebrew speakers show opposite patterns for plurality vs. diminutive contexts.
This is the central result of @cite{berent-bat-el-brentari-dupuis-vaknin-nusbaum-2016}: doubling preferences are not determined by sensorimotor demands (the stimuli are identical novel ASL signs) but by the interaction of morphological context and L1 morphological knowledge.
The dissociation follows from realizeMorphAvailable, which
encodes both positive and negative transfer from L1 morphology.