Debate on No-Root Approach in Linguistics Seminar

Online (hosted by Hebrew U), April 2022 Root Seminar The Pro-Root No-Root debate, with reasons to root for the root Noam Faust Universit Paris 8, CNRS SFL

Outline The No-Root approach a. Melodic Overwriting b. Transfer Effects c. Storage Reasons to root for the root a. Problem of the base: no entry-stem b. Gemination targets c. Neo-Aramaic mono-binyanicity d. Modern Hebrew QoTeT Conclusion: anticipating a counter argument

The No-Root appraoch Best laid-out in Bat-El (1994, 2001, 2002, 2003) and to some extent Ussishkin (1999, 2003). Formally/technically undeveloped since.

No Root Melodic Overwriting Bat El (1994)

No Root Melodic Overwriting Claim in Bat-El (1994): Cluster transfer is not a principle, but a consequence of Melodic Overwriting.

No Root Melodic Overwriting in verbal derivation & Inflection Bat El (2001, 2002, 2003) and to some extent Ussishkin (1999, 2003) reconsider verbal stems in general: PAST B1 B2 B3 gadal grow (intr.) gidel grow (tr.) higdil enlarge

No-Root Melodic Overwriting in verbal derivation & Inflection Bat El (2002, 2003) extends the analysis to inflection steal (tr.) ganav jignov PAST FUT

No-Root Transfer Effects The tendency to preserve aspects of the base when possible: Clusters: fokus => fikes, *fikses faks => fikses, *fikes trol => hitril, *tirel, *tirlel but also avir => ivrer, *iver fa la => fi el,*fi lel berez => hivriz,*birez blof => bilef, *hiblif

No-Root Transfer Effects The tendency to preserve aspects of the base when possible: Vowel: ok, or => okek, orer (but also ikek, irer) rop => la rop (but arap) and possibly also klik => hiklik

No-Root Transfer Effects The tendency to preserve aspects of the base when possible: Vowel: ok, or => okek, orer (but also ikek, irer) rop => la rop (but arap) and possibly also klik => hiklik + Many transfer effects obsereved by Laks in various studies of both Modern Hebrew and palestinian Arabic

No-Root Transfer Effects Claim: Transfer effects are not predicted in a Root Extraction approach This claim relies on the assumption that once the root is extracted, the base form is forgotten . Is that an obligatory assumption of this model though?

No-Root Transfer Effects Recall Bat-El s claim from (1994): Cluster transfer is not a principle, but a consequence of Melodic Overwriting. But even Bat-El claims that the choice of the melody/template is influenced by the form of the base (trol=>hitril; faks=>fikses; fokus=>fikes). So for everybody, there is in fact a general principle: Choose the template that allows maximum maintenence of similarity If that is a principle of any account, why won t it be one guiding the mapping of extracted roots?

No-Root Transfer Effects Recall Bat-El s claim from (1994): Cluster transfer is not a principle, but a consequence of Melodic Overwriting. But even Bat-El claims that the choice of the melody/template is influenced by the form of the base (trol=>hitril; faks=>fikses; fokus=>fikes). So for everybody, there is in fact a general principle: Choose the template that allows maximum maintenence of similarity If that is a principle of any account, why won t it be one guiding the mapping of extracted roots?

No-Root Storage and the problem of the source If [gidel] = [gadal] + <i,e>, then <i,e> is a morpheme. But what about [nigen], given the absence of [nagan]? No-root: [nigen] is an underived word. The entry is /nigen/ (from which [-nagen,nigun] are derived). An entry like /nigen/ wears its binyanic classification on its sleeve

No-Root Storage and the problem of the source An entry like /nigen/ wears its binyanic classification on its sleeve . In other words, it is morphologically complex. Speakers store not /nigen/, but /nigen/. But is /ngn/ also a morpheme? According to No-Rooters, it isn t. Because

No-Root Storage and the problem of the source But is /ngn/ also a morpheme? According to No-Rooters, it isn t. Because i. the appearance of /ngn/ with /i,e/ is unpredictable. ii. /ngn/ never appears alone. iii. Relatedly, the meaning of the combination /nigen/ is unpredictable iii. /ngn/ is unpronounceable. => /ngn/ is a residue.

No-Root Storage and the problem of the source All of these objections are not specific to Semitic morphology: The appearnace of astr with al or naut is unpredictable; astr-al astr-o-nom-y astr-o-log-y astr-o-naut astr-o-man astr-o-dome astr-o-physicist nor is the meaning of the combination. [ str] is not a well-formed word in English. [ str] [ strow] isn t a word. => is it a residue???

No-Root Storage and the problem of the source All of these objections are not specific to Semitic morphology: The appearnace of astr with al or naut is unpredictable; astr-al astr-o-nom-y astr-o-log-y astr-o-naut astr-o-man astr-o-dome astr-o-physicist nor is the meaning of the combination. [ str] is not a well-formed word in English. [ str] [astrow] isn t a word. => The debate is much wider, and I m not sure it can reach a conclusion. But at least we see that Semitic is not special.

No-Root Storage and the problem of the source If the goal of the No-Root approach was to demystify Root & Pattern morphology, then the demystification can go the other way too: All languages have roots. Semitic affixes take the form of templates.

Reasons to root for the root

Reasons to root for the root No Entry Stem in inflection Faust & Hever (2010): Problem of the source in inflection. Within the first binyan of Modern Hebrew, there is no single word-from that distinguishes a paradigm from other paradigms:

Reasons to 4the No Entry Stem in inflection Faust & Hever (2010)

Reasons to 4the No Entry Stem in inflection For every possible entry stem, there are at least two paradigms that need to be distinguished. This cannot be done by applying the same MO rules Faust & Hever (2010)

Reasons to 4the Gemination target In Amharic (Ethiosemistic, Leslaw 1995), the perfect base usually includes gemination of what would be the second root consonant (a). It also occurs in biconsonantal bases (b,c), but not if the preceding vowel is [a,o] (d,e) Elsewhere, geminates can follow [a,o] Perfect Imperfective j s br j s ma j q r j s m j qom Jussive j sb r break j sma j q r j sam j qum rise (a) s bb r (b) s mma (c) q rr (d) sam (e) qom hear remain kiss

Reasons to 4the Gemination target Assume the gemination applies to a base. How does the system know which consonant to geminate? and whether to geminate or not? penultimate surface C? no (c-d) second surface C? Second syllable? no (d,e) Perfect Imperfective j s br j s ma j q r j s m j qom Jussive j sb r break j sma j q r j sam j qum rise (a) s bb r (b) s mma (c) q rr (d) sam (e) qom hear remain kiss

Reasons to 4the Gemination target Assume the gemination applies to a base. How does the system know which consonant to geminate? and whether to geminate or not? penultimate surface C? no (c-d) second surface C? Second syllable? no (d,e) A root-based analysis of (d,e) as sAm, qUm can be proposed. Then: geminate second radical, unless vocalic. Perfect Imperfective j s br j s ma j q r j s m j qom Jussive j sb r j sma j q r j sam j qum (a) s bb r (b) s mma (c) q rr (d) sam (e) qom

Reasons to 4the Neo-Aramaic mono-binyanicity Jewish Urmi Neo-Aramaic (JUNA) (Khan 2008): (roughly) two verbal types

Reasons to 4the Neo-Aramaic mono-binyanicity Jewish Urmi Neo-Aramaic (JUNA) (Khan 2008): (roughly) two verbal types Yet if the first [ ] of T2 is epenthetic, the vocalization is identical in both types => There are no verbal types.

Reasons to 4the Neo-Aramaic mono-binyanicity Recall Bat El s take for underived stems: [nigen] cannot be decomposed because given the root ngn there is no way to predict its vocalization/appearance in B2. => This fact must be stored.

Reasons to 4the Neo-Aramaic mono-binyanicity In Jewish Urmi Neo-Aramaic, given the roots q,t,l and p,r,t,f all stem forms are completely predictable. Since vocalization expresses TAM identically in all verbs, entries are strictly consonantal by economy.

Reasons to 4the Modern Hebrew QoTeT Only with identical final and penultimate Cs (1 exception: roken empty ). Only in B4, B5

Reasons to 4the Modern Hebrew QoTeT Pro-root appraoch. QoTeT forms are parallel to QiTLeL ones The root is reduplicated to fill the 4-place template and associated from right to left. A special allomorphic rule transforms /j/ into /o/ in coda position. The vowel /o/ blocks the regular vocalization of the binyan. Worked out in Faust (to appear); but the idea is old, see references there.

Reasons to 4the Modern Hebrew QoTeT Only with identical final and penultimate Cs. Follows from this being a QiTLeL verbs, with a triradical that needs to be reduplicated in order to satisfy the template Only in B4, B5 Follows from this being a QiTLeL verbs, with 4 radicals only these binyanim allow for such roots.

Reasons to 4the Modern Hebrew QoTeT Only with identical final and penultimate Cs (1 exception). Follows from this being a QiTLeL verbs, with a triradical that needs to be reduplicated in order to satisfy the template Only in B4, B5 Follows from this being a QiTLeL verbs, with 4 radicals only these binyanim allow for such roots. Neither of the two distributional facts follow from a word- based approach

Reasons to 4the Modern Hebrew QoTeT In a word-based approach to QoTeT verbs, there must be an <o,e> pattern: it is used in [ or] => [ orer]. But the pattern must be limited to monosyllabic bases. Otherwise, we predict (b, c) by preservation of the base.

Reasons to 4the Modern Hebrew QoTeT In a word-based approach to QoTeT verbs, there is no way to make the limitation to reduplicated forms follow from the pattern being <o,e>. Reduplication must be assumed to be a requirement of this pattern.

Reasons to 4the Modern Hebrew QoTeT Also, why does the pattern <o,e> occur only in B4,B5? One could develop the following arguement: <o,e> is a subpart of <i,e> because it has an <e> as the second vowel. But then, why are there no QoTaL, noQTaL or hoQTiL verbs? Again, the limitation to B4,5 follows from the root-based account, but has to be assumed by the word-based one. Of course, one could always say this is diachronic. But note that the <o,e> vocalization is alive and well by all accounts, so...

Conclusion The No-root approach offers a refreshing account of Semitic morphology, in an attempt to make it more similar to concatenative systems. It is primarily argued for based on transfer effects. But given that reference to a base form is possible for forms using an extracted root, too, the argument is not compelling (in my view). The storage issue is part of a much larger debate on whether, in a store with pots which all have lids, the lid is meaningful or just a residue .

Conclusion The No-root approach has not taken up any of the challenges levelled against it in the literature, notably (but not only) in the domain of weak roots. Indeed, the technical sides of this approach are extremely under-developed, as its few proponents never tackle issues that are problematic for the approach. This is relevant for

Conclusion The Occam s razor argument Laks (to appear, and in this seminar) says: If Melodic Overwriting can derive all of the necessary forms from other surface forms; and everyone agrees some words are derived from others. then Why do we need root-based derivation?

Conclusion The Occam s razor argument Laks (to appear, and in this seminar) says: If Melodic Overwriting can derive all of the necessary forms from other surface forms; and everyone agrees some words are derived from others. then Why do we need root-based derivation? I hope to have shown that is aBIG IF

Conclusion Argument often heard from Outi I can hear Outi saying: These arguments against the word- based approach concern forms that are weak, form closed groups or are relatively unproductive Such phenomena should not be explained with the same tools as the hyper-productive cases, or they too would be hyper-productive In the same spirit, Laks systematically ignores exceptions to his trends, such as [dafis] printable , dismissing them as non- representative.

Conclusion On my end, and this might be an aesthetic choice, I like problems. I want my theory to allow for any regularity, even those that are only mildly or marginally productive. To this end, I think that the notion of the root is necessary.

Thank You!

References Bat-El, Outi. 1994. Stem Modification and Cluster Transfer in Modern Hebrew. Natural Language & Linguistic Theory 12:4, 571 96 Bat-El, Outi. 2002. Semitic verb structure within a universal perspective. In J. Shimron (ed.), Language Processing and Acquisition in Languages of Semitic, Root-based, Morphology, 29-59. Amsterdam: John Benjamins. Bat-El, Outi. 2003. The Fate of the Consonantal Root and the Binyan in Optimality Theory. Recherches Linguistiques de Vincennes 32, 31 60 Faust, Noam. 2019. New reasons to root for the Semitic root from Mehri and Neo-Aramaic . The Linguistic Review 36 (3), 575-599. Faust, Noam. To appear. QoTeT verbs in Modern Hebrew and the Pro-Root vs. No-root debate. Faust, Noam and Yaar Hever. 2010. Empirical and Theoretical Arguments for the Discontinuous Root in Semitic. Brill s Annual of Afroasiatic Languages and Linguistics 2: 80-118. Khan, Geoffrey. 2008. The Jewish Neo-Aramaic Dialect of Urmi. Piscataway, NJ: Gorgias. Leslau, Wolf. 1995. ReferenceGrammar of Amharic. Weissbaden: Harrassowitz. Lowenstamm, Jean. 1996. CV as the only syllable type. In Current trends in Phonology. Models and Methods, ed. Jacques Durand & Bernard Laks. Salford, Manchester: ESRI. 419-441 Prunet, Jean-Fran ois. 1996. Guttural vowels. In Essays on Gurage Language and Culture, ed. by Grover Hudson, 175 203. Wiesbaden: Harrassowitz. Ussishkin, A. 1999. The Inadequacy of the Consonantal Root: Modern Hebrew Denominal Verbs and Output output Correspondence , Phonology 16, 401 42 Ussishkin, Adam. 2003. Templatic Effects as Fixed Prosody: The Verbal System in Semitic. In Research in Afroasiatic Grammar, vol. 2 edited by Jacqueline Lecarme, 511 530. Amsterdam: John Benjamins