Conference Abstracts

Friday June 1, 2018

The functional load of tone in Hausa
Nicholas Rolle (UC Berkeley), Stephanie Shih (USC) and Sharon Inkelas (UC Berkeley)

1 Tone in Afroasiatic
Contrastive tone is found throughout two branches of Afroasiatic: Chadic and Cushitic. For Chadic, Schuh (2017:91) notes that ‘tone has to be overtly specified for each and every noun’ which ‘shows it to be an important, phonological component of the lexicon’. However, he also notes that exact minimal pairs are uncommon’ (e.g. /L/ vs. /H/ in Margi [mrt], fà‘farm’ vs. fá ‘year’ - Schuh 2017:93). Table 1 shows the low number of nouns distinguished solely by tone.

Lang ISO n = (approx.) Distinguished by tone
Bole bol 2000 139 6.95%





















Table 1: Low number of tonal minimal pairs in Chadic (interpreted from Schuh 2017:90)

Where tone plays a small role in distinguishing morphemes, tone is said to have a low functional load, claimed for Chadic languages Makary Kotoko [mpi] (Allison 2012:38) and Goemai [ank] (Tabain & Hellwig 2015:91), and for Cushitic as a whole (Mous 2009), e.g. Awngi [awn] (Joswig 2010:23-24). However, most statements do not provide clear statistics to support their position.

2 Quantifying the functional load of tone
In this talk, we present the first study quantifying the functional load (FL) of tone in Hausa. While there is a long history measuring FL in phonology (Martinet 1955, 1977; Hockett 1967; Wedel et al. 2013a,b; Oh et al. 2013, 2015), very few studies have looked at tone. Our study employs a methodology measuring entropy (H), an information theoretic term meaning the amount of unpredictability in a system. From a digitized Hausa lexicon of 10,768 lexemes, we assess the FL of four phonological properties: (1) consonant quality, (2) vowel quality, (3) tone, and (4) vowel length. We do this by quantifying a baseline entropy for the lexicon (HB), then assess the change in entropy if we hypothetically merge phonological contrasts (e.g. merging consonant quality: dala, tala, mana > XaXa). The FL is the amount of uncertainty (=entropy) that is lost by the merger (Hall et al. 2016).

Doing this for all contrast mergers, and calculating it for all lexemes, vs. for nouns, verbs, and graded verbs, results in a set of FL measurements for the lexicon, shown below. One can see that while the FL for consonants is high (FLC = .103), it is low for vowels (FLV = .013), and half that value for tones (.006).

Unique entries  before merger Types of hypothetical mergers testing FL
No C Qual No V Qual No Tone No Vowel Length (VL)








HB = 13.357

HNoC = 11.975 

HNoV = 13.177

HNoT = 13.278

HNoVL = 13.326

FLC = .103

FLV =.013

FLT =.006

FLVL =.002








HB = 12.702

HNoC = 11.617

HNoV = 12.604

HNoT = 12.674

HNoVL = 12.693

FLC = .085

FLV =.008

FLT =.002

FLVL =.001








HB = 11.239

HNoC = 8.983

HNoV = 11.023

HNoT = 11.151

HNoVL = 11.198

FLC = .201

FLV =.019

FLT =.008

FLVL =.004

verb grade (n=2452)






HB = 11.240

HNoC = 9.047

HNoV = 11.166

HNoT = 11.240

HNoVL = 11.220

FLC = .195

FLV =.007

FLT =0

FLVL =.002

Table 2: Assessing the functional load of tone in Hausa

3 Discussion
In contrast to Hausa, previous research in tonal languages Mandarin and Cantonese has shown that while the FL of consonants is similarly much higher compared to vowels, the FL of vowels is largely equivalent to the FL of tone, showing their equal lexical importance (Surendran & Niyogi 2003, 2006; Surendran & Levow 2004; Oh et al. 2013, 2015). We put forward the Functional Load of Tone (FLT) hypothesis, which states that in languages where the functional load of lexical tone is low (e.g. Hausa), the use of grammatical tone is relatively high, and vice versa. This hypothesis is supported by the large role grammatical tone plays in Afroasiatic tone systems (Mous 2009, Schuh 2017), and its diminished role in these Chinese languages.

A new perspective on Mehri diminutives
Morgan Rood (Carleton College)

Utilizing data from fieldwork, this project proposes a new analysis of Mehri diminutives that relies primarily on phoneme insertion and phonological processing rather than the CV templates of more traditional Semitic research. Diminutives in Mehri (Modern South Arabian) were first thoroughly documented by Johnstone (1973). Utilizing the Semitic notation of rootand- pattern morphology, Johnstone (1973) develops three categories for diminutive formation: Cewɛ̄CēC, CeCɛ̄Cēn, CeCɛ̄CēC. In her grammar, Watson (2012) describes a much more robust system (11 patterns overall). These forms are more inclusive of feminine and plural morphology that that had been simplified in the Johnstone typology.

My analysis of nominal diminutives builds upon the generalizations from previous research and develops a more unified account. Minimally, a diminutive singular noun is formed with the infixation of –ā- at the penultimate syllable. For example:

(1) a.  hūlaʕ           b.  hw<ā>laʕ
        shadow.M            shadow.M<DIM>
(2) a.  θḥm-ōt          b.  θḥ<ā>m-ōt
        cinder/ash-F        cinder/ash<DIM>-F
(3) a.  mɬ̣ġ-ōt          b.  mɬ̣ġ<ān>-ōt
        bite/morsel-F       bite/morsel-F
These diminutive forms can be formed from either a feminine or masculine noun. Note that in (3), which ends in the feminine suffix –Vt, an epenthetic –n- is inserted following the –ā- to avoid hiatus.1

In some nouns, the diminutive infix appears in conjunction with a suffixed –ān:

(4) a.  ḥmūh            b.  ḥm<ā>h-ān2
        water.M             water.M<DIM>-DIM
The diminutives seen in (4), which Johnstone (1973) characterized as “Type 2”, all end in - ā<last radical>ān. I argue against the analysis that these diminutives are of a different type. Instead, the suffix -ān is a result of a phonological process to ensure a well-formed word. Infixed –ā- is not licensed in either the first syllable or the final syllable (inclusive of suffixes). In other words, -ā- is only licensed within a minimally trisyllabic word. When these restrictions preclude the insertion of –ā- in any position, -ān is suffixed to the word to “save” the derivation.

The diminutive plural noun consists of the infixed –ā- and the suffix –ūtan. Unlike the diminutive singular, where the –ā- is found on the penultimate syllable, the infixed –ā- for the plural diminutive is found on the antepenultimate syllable:

(5) a. kabkīb           b.  kabk<ā>b-ūtan
    star.M                  star<DIM>-PL.DIM
(6) a. raḥb-ēt          b.  raḥb<ān>-ūtan
    city-F                  city<DIM>-PL.DIM
Seen in (6), epenthetic –n- can also be inserted to avoid hiatus in diminutive plurals.

The diminutive data presented in this project, including dozens of novel forms, is the most comprehensive contribution to the diminutive descriptive literature since Johnstone (1973). The present analysis is summarized in the table below.

Present Analysis Template from Johnstone (1973) Template from Watson (2012) Examples from new data

Insertion of –ā- in penultimate syllable 

hw<ā>laʕ (1)
θḥ<ā>m-ōt (2)
Insertion of –ān- in penultimate syllable CaCCānōt mɬ̣ġ<ān>-ōt (3)
Insertion of –ā- + suffixed –ān CeCɛ̄Cēn C(a)CāCān ḥm<ā>h-ān (4)
Insertion of –ā- + plural suffix -ūtan CaCCāCūtan;
kabk<ā>b-ūtan (5) 
Insertion of –ān- + plural suffix -ūtan   CaCCānūtan raḥb<ān>-ūtan (6)

1 Consistent with Mehri phonology. Mehri employs several strategies to avoid hiatus (Watson 2012).
Consonant harmony in Tunisian Arabic: could it be a contact outcome with Tamazight
Soubeika Bahri (CUNY)

Consonant harmony (CH), defined as assimilation between non-adjacent consonants (Goad 2004, Hanson 2010), is observed in the phonology of contemporary Tunisian Arabic and manifests in a regressive directionality. Voiced alveolar fricatives trigger assimilation to preceding voiced post-alveolar fricatives’ place of articulation as in the following examples:

Arabic with no CHTunisian Arabic with CH
a. ʒazzarzazzar ‘butcher’
b. ʒlizzliz ‘tiles’
c. mʒæzmzæz ‘a small entrance’
d. ʒɪzzazɪzza ‘sheep wool’
e. ʒaziazazɪa ‘feminine proper name’
f. ʕaʒuzʕazuz ‘old man’
g. ʒhazzhaz ‘dowry’

Several accounts attribute the distribution and use of consonant harmony feature to segmental, phonotactic, and prosodic language-internal motivations (Berg 1992, Vihman 1978, 2007). While language-internal motivations are not totally ruled out, this study argues that consonant harmony in Tunisian Arabic is a product of language contact with Tamazight, the indigenous language of the region, particularly that similar feature is commonly attested in several Tamazight dialects/languages in North Africa and in other Arabic dialects spoken in the same region (Hanson 2004, Zellou 2010, Kossman 2013).

The study uses data collected through a prompted production task of elicited imitation (Patrick & Mackey 2013). At first, 40 sentences with words where the sibilant regressive assimilation is expected to happen were presented orally to 10 Tunisian adults who do not speak Tamazight. The same sentences were presented later to 10 Tunisian adults who speak both TA and Tamazight. All the participants had to listen to the sentences then repeat them accurately when prompted to do so. All the data was recorded. A total of 800 tokens were transcribed using the IPA system, then analyzed. Results show that all the participants assimilate their voiced post-alveolar fricatives to voiced alveolar fricatives when found in the same word and when with the former feature is found in an environment that precedes the latter.

To demonstrate how language contact is responsible for the occurrence of consonant harmony in TA, this study makes use of contact-induced language change framework as proposed by Thomason and Kaufman (1988) and Thomason (2001). This study concludes that long distance language assimilation in Tunisian Arabic is inherited through contact-induced change without excluding the potential that both, the external and the internal hypotheses, are responsible for the emergence of this type of consonant harmony in TA.

Effects of dialect, gender and context on vowel durations in Oromo
Feda Negesse (Addis Ababa University) and Tujube Amansa (Addis Ababa University)

An acoustic study of vowels of Cushitic languages has not been conducted though their vowels seem to exhibit durational variations due to phonetic and phonological reasons. Oromo is one of the Cushitic languages which has the largest number of speakers. We investigate effects of regional dialect, gender, and phonetic context on vowel durations of Oromo. Sixty-four speakers (F= 32, M= 32) from five dialect areas of the language produced vowel data in different contexts. An acoustic analysis of the data shows that durations of the vowels significantly vary across dialects, with the longest durations in the Southern Eastern dialect and the shortest durations in the Northern dialect. Gender has no significant effect on durations of vowels but both long and short vowels have significantly longer durations when followed by voiced singletons than voiceless, and when followed by voiced singletons than voiced geminates. Collectively, longer vowels are two times longer than the short ones and this difference is significantly affected by dialect and context but not by gender. It is concluded that dialectal background of speakers and phonetic context are key factors determining durations of Oromo vowels.

The domain of emphasis spread in Arabic: Evidence from urban Jordanian Arabic
Aziz Jaber (Yarmouk University), Osama Omari (Yarmouk University) and Rasheed Al-Jarrah (Yarmouk University)

Several studies, primarily phonological, have shown that emphasis in Arabic spreads from the emphatic sound to its neighboring segments (Broselow 1976, Younes 1982; 1993, Davis 1995, Watson 1999, Zawaydeh 1999, Jaber 2001, Bin-Muqbil 2006, Jongman et al. 2011, among many others). These studies reveal that Arabic dialects show differences in the phonological domain of emphasis spread. According to Broselow (1976), Jaber (2001), and Jongman et. al (2011), the domain of emphasis spread is the syllable, but to Watson (1999), Zawaydeh (1999), and Al- Khatib (2008) it is the entire word, and to Younes (1982; 1993), it is the uninflected word. We strongly believe that one reason why previous research on emphasis spreading still produces conflicting findings is that these studies still assume an immediate correspondence between the phonetics of emphatic sounds and their phonological realization, relegating the role of the morphological make-up of the word. Furthermore, most previous studies on emphasis spreading have used the Arabic trigging emphatic sounds (whether primary or secondary) indiscriminately, yet making generalizations that they think are applicable to all of them.

To demonstrate a fuller account of the domain of emphasis spreading in Arabic, in general, and Urban Jordanian Arabic (UJA), in particular, the current study provides an acoustic analysis of emphasis spreading in polysyllabic words (monomorphemic vs. polymorphemic) as produced by ten native speakers of UJA (5 males; 5 females). The stimuli consisted of a list of (56 words) minimal pairs containing the emphatic sound /tˤ/ and its plain counterpart /t/ in word-initial and word-final position. Vowel formant frequencies (F1-F3) were measured at the midpoint of the vowels preceding or following the target sound.

To this end, we hope to show that the morphology of the word is a real confounding force that constrains the operation of emphasis spreading on the entire word. The point of departure from probably all previous work on emphasis spread is the assumption that emphasis is not merely a phonological but a morphophonological component.

The findings of the current study, F2 results in particular as F1 and F3 prove to be slightly irrelevant to the identification of the domain of emphasis in UJA, show the effects of morphological complexity in identifying the nature and domain of emphasis spreading in this dialect. The F2 of all the vowels in polysyllabic monomorphemic words was significantly lowered (significant by a 1-tailed t-test, p < 0.0000). However, the F2 of the vowels in the neighboring morphemes in polymorphemic minimal pairs did not significantly change in the presence of the emphatic consonant (not significant by a 1-tailed t-test, p=0.741 for first syllable, and p= 0.828 for final syllable). The results also show that there is a significant difference in F2 measurement in monomorphemic emphatic words depending on the position of the target emphatic sound (/tˤ/) within the word. The mean difference of F2 between emphatic and non-emphatic monomorphemic words with the emphatic/plain sound in the first syllable (156.886 Hz.) was lower than that of the syllable with the emphatic/plain consonant in the final syllable of the word (215.357 Hz.). More evidently, in trisyllabic monomorphemic words containing the emphatic consonant, F2 in the distant vowels was significantly lowered as opposed to those containing the plain counterpart. However, no significant difference in F2 measurement in the distant vowels in trisyllabic polymorphemic words has been attested. This apparently indicates that spread is a property of the morpheme. The results of the current study contradict the older 'distance from trigger' model.

2:00Incomplete phonetic neutralization in Arabic vowel alternations
Nancy Hall (California State University, Long Beach)
An acoustic analysis of Amharic fricatives
Derib Ado Jekale (Addis Ababa University)

This study presents an acoustic analysis of Amharic, an Ethio-Semitic language spoken by over 20 million native speakers and the working language of the Federal Government of the Federal Democratic Republic of Ethiopia. The language has five voiceless fricative phonemes: /f, s, s’, ʃ, h/. Data for the study was collected from 5 male and 5 female speakers aged 20-32 and born and raised in Addis Ababa. A Sennheiser e-815 dynamic microphone was used to record audio data on CSL 4400 attached to a computer with a sampling rate of 44100 Hz and quantization of 16 bits in a quite room. All the fricatives were put in disyllabic Amharic words that begin with a fricative followed by the vowel /a/. Duration of noise, spectral moments (spectral mean, skewness, kurtosis and spectral variance), zero crossing rate and intensity) measurements were taken using Praat and analysed. The results showed that the alveopalatal fricative had the longest mean duration followed by the alveolar fricative, the labio-velar fricative and the glottal fricative. Female participants produced longer fricative sounds than males did at all place of articulations. Among spectral measurements, spectral mean, kurtosis and spectral standard deviation were found to have effects on place of articulation, but the patterns were the same for male and female participants for spectral mean and mean intensity only. The effect of airstream was seen clearly with the ejective fricative /s’/ having shorter duration, and higher spectral mean, kurtosis and standard deviation than the alveolar fricative /s/ for both male and female speakers.

On the case system of Berber
Lydia Felice (Georgetown)

A contested topic among linguists is the purpose of the distinction between the Free State (FS) and the Construct State (CS) in Berber (see e.g. Mettouchi 2015; Guerssel 1992; Ennaji 2001). In Kabyle Berber, nominals may appear in the FS (1a, 2a); or CS (1b, 2b). FS nominals are characterized by the presence of the prefix a-. CS nominals lack this prefix; feminine CS nouns lack any additional prefix, as in (2b), and masculine CS nouns are prefixed by the morpheme w- (1b).

(1a) a-qcic (1b) wqcic (2a) t-a-qcic-t (2b) t-qcic-t
    FS-child child.CS F-FS-child-F.Sg F-CS.child-F.Sg
    ‘boy ‘boy’ ‘girl’ ‘girl’

I propose that the prefix a- is an intrinsic case licenser that occupies K0 and licenses Case to the nominal, much like the augment vowel in Zulu (Halpert 2015). This proposal accounts for the full distribution of FS and CS nominals and has implications for our understanding of Case.

Nominals in the FS appear as (i) preverbal subjects, (ii) objects of a verb, and (iii) complements of certain so-called prepositions. Nominals in the CS appear as (i) postverbal subjects and (ii) complements of other prepositions. The distributions of the FS and CS nominals do not form obvious natural classes; as illustrated in (5) and (6), subjects may appear in the FS or CS, depending on where they appear with respect to the verb. In addition, while some prepositions select FS complements (7), others select CS complements (8).

(5) A-rgaz y-ecca. (6) Y-ecca wrgaz. (7) al ajdir (8) γer wjdir
    FS-man 3M.Sg-ate 3M.Sg-eat man.CS until cliff.CS to cliff
    ‘The man has eaten.’ ‘The man has eaten.’ ‘up to the cliff’ ‘to the cliff’

The distribution of the FS and the CS suggest that the distinction is syntactically conditioned. I assume, following Guerssel (1992), that the FS morpheme is a case marker occupying K0. Building on Guerssel’s insights, as well as Halpert’s analysis of augmented nominals in Zulu, I propose a new account of the FS/CS distinction as a difference in the size of the nominal projection. Nominals in the CS are bare DPs (contra Guerssel 1992) that must receive structural Case from elsewhere in the clause, while nominals in the FS are KPs that receive Case from an intrinsic licenser, the FS morpheme a-. I propose that this analysis accounts for the full distribution of Free State and Construct State nominals.

An outstanding problem is that preverbal subjects appear in the FS, while postverbal subjects appear in the CS. I assume, following Shlonsky (1987), that preverbal subjects are base-generated clause externally and coindexed with a pronoun in the (postverbal) subject position. Berber is a pro-drop language, so this pronoun is null when the subject is dislocated. I bring new evidence from Kabyle to bear on this claim. Postverbal subjects receive structural case in their base positions, and thus surface in the CS. Preverbal subjects are generated in a position where structural case is unavailable. To avoid a case filter violation, they must be licensed via an inserted K0, the FS vowel a-.

The interaction of discontinuous agreement with auxiliaries in Amharic: A distributed morphology approach
Ruth Kramer (Georgetown)

In Amharic (Semitic; Ethiopia), subject agreement is discontinuous, i.e., it can be expressed by both a prefix and a suffix. Discontinuous agreement across languages has been the focus of considerable morphosyntactic research (e.g., Noyer 1992, Halle 1997, Trommer 2002, Harbour 2007, 2008ab, Campbell 2012). However, most of this work has focused on verbs in isolation. In this paper, I examine the interaction of discontinuous agreement with a following auxiliary in Amharic. I develop a novel analysis using the Distributed Morphology operation Local Dislocation, and demonstrate how this analysis fares better than alternative approaches.

In Amharic, imperfective verbs can combine with the non-past-tense suffix/enclitic auxiliary -all (this auxiliary also displays subject agreement; subject agreement is glossed as .S, object clitics as .O).

(1) tɨ- säbr- iy- all- äʃ
    2.S- break.IPFV- 2FSG.S- AUX- 2FSG.S
    ‘you ( break, you ( will break’ (Leslau 1995:342)
However, when the auxiliary is present, the plural subject agreement suffix –u ((2)a) is absent ((2)b).
(2) a. yɨ- säbr- u ‘they break’
    3.S- break.IPFV-PL.S
    b. yɨ- säbr- all- u ‘they break, they will break ’
    3.S- break.IPFV-AUX-PL.S (Leslau 1995: 301, 342)
I argue that this is a low-level, morphological effect. As seen in (1), other agreement suffixes are present when followed by this auxiliary. Moreover, the plural suffix remains present when followed by an different, free-standing auxiliary or by a different kind of suffix (e.g., negation; Leslau 1995: 316, 303).

Adopting Distributed Morphology, I develop an analysis of (2)b using the post-syntactic, morphological operation Local Dislocation (Embick and Noyer 2001). Local Dislocation occurs after linearization and switches the position of two adjacent elements. The underlying order of (2)b is therefore like (1), but the plural subject agreement suffix and the auxiliary ‘flip’ positions ((3)).

(3) Underlying order: [yɨ * nägr * u * all]             →           [yɨ-nägr-all-u]
                                                LOCAL DISLOCATION
Object clitics provide further support for this approach. They come between subject agreement suffixes and auxiliaries, and when one is present, the plural subject agreement suffix appears on the verb and the auxiliary lacks –u.
(4) yɨ- nägr- u- ññ- all
    3.S- tell.IPFV- PL.S- 1SG.O- AUX
    ‘they will tell me, they are telling me’ (Leslau 1995:421)

This is predicted by a Local Dislocation analysis. In (4), since the plural subject agreement suffix and the auxiliary are not adjacent, Local Dislocation is not triggered and the plural subject agreement suffix surfaces next to the verb.

The paper concludes with some discussion of alternative approaches. For example, (2)b could be explained by contextual allomorphy, i.e., plural subject agreement having a null allomorph in the context of a following auxiliary. However, a contextual allomorphy approach would struggle to explain the “disappearance” of –u from the auxiliary in (4). Additionally, the analysis in this paper points towards treating discontinuous agreement as morphological, not syntactic (pace Halefom 1994 on Amharic, Banksira 2000 on Chaha); I show how previously-discussed evidence in favor of a syntactic approach is no longer problematic assuming Distributed Morphology.

On the syntax of ellipsis: Sluicing and cleft-sluicing in Egyptian Arabic
Usama Soltan (Middlebury College)

Sluicing is an elliptical structure that has been claimed to result from wh-fronting and deletion (Ross 1969; Merchant 2001), as in (1).

1) John bought something, but I don’t know [what [John bought]]

Egyptian Arabic (EA), a wh-in-situ language that also allows wh-clefting and wh-fronting under specific conditions, provides support for this correlation between wh-fronting and sluicing. More specifically, this paper argues that EA exhibits genuine sluicing only when wh-fronting is allowed, but cleft-sluicing otherwise. The paper thus contributes to the cross-linguistic study of ellipsis and its syntactic analysis.

EA utilizes multiple strategies to form wh-questions (Wahba 1984). Wh-arguments appear in situ (2), or as pivots of a cleft structure (3), but never fronted (4).

2) shuf-t miin?      3) miini (huwwa) (da) Ɂillii Ɂinta shuf-t-uhi?     4) *miini shuf-t?
   saw-2SGM who         who COP.SGM DEM.SGM COMP you saw-2SGM-him           who saw-2SGM
   ‘Who did you see?’   ‘Who is it that you saw?’

Bare wh-adjuncts appear in situ (5), or fronted (6); but never as pivots of a cleft structure (7).

5) ha-tsaafir feen/Ɂimtaa/Ɂizzaay/leeh?       6) feen/Ɂimtaa/Ɂizzaay/leeh ha-tsaafir?
   FUT-travel.2SGM where/when/how/why?           where/when/how/why FUT-travel.2SGM
   ‘Where/when/how/why will you travel?’         ‘Where/when/how/why will you travel?’
7) *feen/Ɂimtaa/Ɂizzaay/leeh (huwwa) Ɂillii ha-tsaafir?
    where/when/how/why COP.SGM COMP FUT-travel.2SGM
    ‘Where/when/how/why is it that you will travel?’

If the correlation between wh-fronting and sluicing holds, we expect EA wh-adjuncts, but not wharguments, to appear in sluicing contexts. This is not borne out, however, since both (8) and (9) are equally grammatical.

8) Mona ha-tsaafir bass ma-ʕraf-š feen/Ɂimtaa/Ɂizzaay/leeh
   Mona FUT-travel.3SGF but NEG-know.1SG-NEG where/when/how/why
   ‘Mona will travel, but I don’t know where/when/how/why.’
9) Mona bi-tiħibb waaħid bass ma-ʕraf-š miin
   Mona ASP-love.3SGF one but NEG-know.1SG-NEG who
   ‘Mona loves someone, but I don’t know who.’

Since wh-arguments cannot be fronted (4), (9) is unexpected. There is evidence, however, that structures like (9) are sluicing-like constructions (SLCs), resulting from a wh-cleft structure that undergoes deletion. Under this cleft-sluicing analysis, the wh-phrase is a pivot of a cleft structure, followed by deletion of the rest of the clause under semantic identity with the antecedent clause, as schematically represented in (10); cf. Eid 1983; Ouhalla 1999; Choueiri 2016.

10) [Cleft_Clause miini [Copular_Clause Subjecti Copula Ɂillii Mona bi-tiħibb-uhi]]

There are several arguments for this analysis. First, like wh-cleft structures (3), SLCs may surface with a pronominal copula and/or a demonstrative (11).

11) Mona bi-tiħibb waaħid bass ma-ʕraf-š miin (huwwa) (da)
    Mona ASP-love.3SGF one but NEG-know.1SG-NEG who COP.SGM DEM.SGM
    ‘Mona loves someone, but I don’t know who.’

Second, as in clefts, either the copula or demonstrative receives the pitch accent in SLCs. Third, the analysis correctly predicts that wh-PPs appear in both sluicing and cleft-sluicing (12).

12) Mona bi-titkallim maʕa waaħid bass ma-ʕraf-š miin / maʕa miin
    Mona ASP-talk.3SGF with one but NEG-know.1SG-NEG who / with who
    ‘Mona is talking with someone, but I don’t know who/with whom.’

In sum, EA exhibits genuine sluicing only when wh-fronting is permitted, and cleft-sluicing otherwise, providing evidence for the correlation between sluicing and the wh-syntax of the language.

Saturday, June 2

Language change in Amharic
Zelealem Temesgen (Addis Ababa University)

This presentation discusses the phonological, lexical, syntactic and semantic changes witnessed in present-day Amharic. The qualitative data collected from spoken and written sources suggest that sounds are being articulated differently. The replacement of ejectives by non-ejectives as in ኢትዮጵያ /itoyop'oya/ 'Ethiopia' vs. ኢትዮፕያ /itoyopoya/ has become common among the younggeneration speakers. Ease of articulation or the tendency to be viewed as English-speaking, and hence 'modern' appear to account for this phonetic change. The tap /r/ is being articulated trilled. The 3MSG marker /-h/ is being replaced by /-k/ as in ደህና ነህ /dähna näh/ vs. ደህና ነክ /dähna näk/ 'How are you 2MSG?'. The young speakers of Amharic use the accompaniment marker ጋር /gar/ instead of the locative marker ጋ /ga/ as in አንተ እዛ ጋር ቁም /antä ozza gar k'um/ instead of አንተ እዛ ጋ ቁም /antä ozza ga k'um/ to mean 'You (2MSG) stand over there!'. The second person polite form is being replaced by the third person plural form as in ምን ትፈልጋላችሁ? /mon tofällogalla––ohu/ 'What do you (2PL) want?' instead of ምን ይፈልጋሉ? /mon yofällogallu/ 'What do you (2RES.) want?'. This phenomenon is undoubtedly the result of the influence of Afan Oromo which uses the plural forms of pronouns for respect. There is now a tendency to replace the 1SG marker /-hu/ by /-w/ in verbs as in ነቅቻለሁ /näk'-ï––-all-ähu/ vs. ነቅቻለው /näk'-ï––-alläw/ 'I am well aware.'. Double accusative case marking as in የእንግዶችን ማሳረፊያን ይመለከታል፡፡ /yaïngïdo––- ïn maräfiya-n yïmmäläkkät-all/ 'It concerns the resting place of guests.' has become common. Tense incompatibility as in ትናንት ማክሰኞ ነው፡፡ /tïnantïna maksäììo näw/ *'Yesterday is Tuesday.' instead of ትናንት ማክሰኞ ነበር፡፡ /tïnant maksäììo näbbär/ 'Yesterday was Tuesday.' has become customary. The extended greeting system is being reduced only to ሰላም /sälam/ 'peace'- based greetings.

Quite a large number of words are added into the Amharic lexicon either through borrowing, compounding or semantic extension. New expressions are introduced from English through loan translation as in ግማሽ ወንድም /gommaS wondomm/ 'half brother', ወንበር ውሰድ! /wonbär wusäd!/ 'Take a chair!', በአንድም ሆነ በሌላ መንገድ /bä-andomm honä bälela mängäd/ 'in one way or another' and so on. እንደ /ondä/, which functions as a subordinator and a preposition, has now added an adverbial function. በዋናነት /bä-wanna-nnät/ 'mainly' is a new frequently used emphatic construction. New trends of subordination have emerged in spoken and written Amharic. The word order has become loose subject to idiosyncrasies. All the changes witnessed are internally or externally motivated. The data apparently show that Amharic is changing rapidly. Quite a significant number of speakers believe that the language is developing, but at the same time worry, that it is suffering from unspeakable linguistic turbulence.

Revisiting the independent pronouns in Hamar: a historical-comparative perspective
Moges Yigezu (Addis Ababa University)

Independent pronouns in Hamar have been briefly described in previous studies such as Lydall (1976), Moges (2005) and Petrollino (2016). In the historical-comparative studies made on Hamar and related languages of the Ariod group, the origin of these pronouns has been a point of discussion due to their resemblance to some neighboring Nilotic languages such as Nyangatom.

Bender (2000:198), for instance, assumed that the third person singular forms of Hamar (kisi and kosi) must be a loan from Nilotic. Zaboriski (2004:181) rejects the idea of borrowing and argues that these pronouns prove that Aroid languages are in fact Nilotic in their origin. Moges (2016) has shown the similarity of the same independent pronouns to the Nilo-Saharan Surmic further suggesting that the similarity of these independent pronouns goes beyond Nilotic.

The current contribution makes further investigation into the independent pronouns in Hamar from a historical-comparative perspective and tries to shed some light on the possible origins of these pronouns.

10:00The development of subject case marking in Omotic-Mao
Michael Ahland (California State University, Long Beach)
Overt/null subjects in Algerian Arabic: A quantitative approach
Osama Omari (Yarmouk University) and Amina Madani (Yarmouk University)

Although the pro-drop phenomenon has received much attention in the previous literature of Arabic (Eid 1983, Parkinson 1987, Moutaouakil 1989, Schulte-Nafeh 2005), only a few studies have examined this phenomenon using quantitative methods (Owens et al 2010, Omari 2016, Al-Shawashreh 2016). The present study aims to fill in this gap in the literature by investigating the potential influence of a number of sociolinguistic factors on the choice of subject form (null/overt) in Algerian Arabic using the variationist framework (Labov 1972). The factors coded in the study include co-referentiality, topicality, tense form, grammatical person and number, transitivity, semantic class of the verb, animacy, clause type, gender and the bilinguality level (Arabic & French).

The data for this study are based on a spontaneous speech of 22 bilingual speakers obtained through recorded interviews. All the participants acquired Algerian Arabic as their first language; however, they differed in the age period they acquired/learned French. Accordingly, the participants are stratified into two social groups: early and late bilinguals. Early bilinguals (8 females; 3 males) are identified as those who acquired French at home before going to school. The late bilingual group (8 females; 3 males), on the other hand, did not acquire/speak French at home. They started learning French as a school subject at the age of 10.

A total of 2139 tokens were extracted and then analysed in Goldvarb. The multivariate analysis of the data shows that the choice of subject form in Algerian Arabic is constrained by co- referentiality, topicality, tense form, person, the semantic class of the verb. However, none of the extra-linguistic factors was selected as significant for determining subject form. These results reveal that external- social factors do not play a role on subject choice. Only linguistic factors are found to constrain the choice of subject form. Algerian speakers under the influence of L2 French do not seem to undergo attrition, thus encouraging further studies questioning the interface hypothesis (Tsimpli et al. 2006).

Coptic noun class and the morphophonology of gender
Chris Reintges (CNRS/LLF/University Paris 7) and Sabrina Bendjaballah (Laboratoire de Linguistique de Nantes)

§1. Background. There exists a long-standing tradition of research on Coptic [Afroasiatic; 3rd–12th century CE] noun classes, dating back to Steindorff’s (1884) dissertation. The focus of scholarly attention has been on the internal reconstruction of nominal(ized) forms in earlier stages. Due to the diachronic bias, the synchronic description of the nominal classification system is far from being complete.

§2. Major Claims. Akin to its Afroasiatic neighbors, Coptic has a two-term, masculine–feminine system. The traditional wisdom within the field is that gender has no morphological exponent on the nominal stem itself and must accordingly be inferred from gender agreement with prenominal definite determiners or anaphorically related pronouns (Layton 2000: 85 §105(a)). Revisiting the issue, we shall argue instead that masculine- and feminine-gendered nouns are distinguished on a morphophonological basis.

§3. Sex-declinable nouns. We shall first turn to masculine–feminine pairs of sex differentiated noun lexemes. As shown in Table 1, same-root sex declinable nouns fall into two classes. In the first class, feminine nouns are marked inflectionally by adding to the nominal stem a vowel suffix –e, which derives from the common Proto-Afroasiatic feminine suffix=–í=(Kramer 2015: 18–19). In the second class, masculine and feminine nouns are differentiated by distinct vocalisms. Vowel alternations can also be found in gendered-suffixed nouns of the first class.

Table 1. Same-root sex-declinable nouns

§4. Homophonous root nouns. The two classes reappear in biliteral nouns derived from homophonous roots. In Class I, masculine nouns have a C1VC2 template with a short central vowel /~/. The gendersuffixed C1VVC2É template of the corresponding feminine nouns contain either a lengthened high or a mid-low front vowel /iː/ and /ɛː/.

Table 2. Masculine and feminine-gendered nouns derived from homophonous biliteral roots

§5. Conclusions. Although the vowel correspondences appear to be systematic, masculine- and feminine-gendered nouns formed with same or homophonous roots are not directly correlated through apophony. In this way, the Coptic noun class system defies straightforward classification in gender typology (Corbett 1991). It rather looks as if we are dealing with a noun class system, in which the obsolete status of gender inflection is compensated for by regular patterns of vowel correspondences and distributional restrictions.

On the syntax and semantics of the transitivity of prepositional verbs in Arabic
Basem Al-Raba'a (Oberlin College)

The notion of transitivity, which is well attested in natural language, is normally understood not only as a property of the verb but also of the entire clause (Hopper and Thompson, 1980, p. 294). The characterization of transitivity is generally, but not necessarily, based on parallel syntactic and semantic grounds. A clause is considered transitive if it involves a nominative subject and an accusative direct object; the former is a volitional, actively-instigating Agent, whereas the latter an inactive, affected Theme (see Lyons, 1968, Hopper and Thompson, 1980, Givón, 1995, and Kittilä, 2002). In traditional Arabic grammar, constructions may be classified as transitive if they involve verbs that assign accusative Case to their direct or cognate object (see Sibawayh, 1316/1898, Levin, 1979, Saad, 1979) or only to their direct object that is also a Theme argument (Ibn Jinni (as cited in Schapiro, 2000)). The delineation of transitivity, however, extends beyond the subject-direct object relation. This is manifested in Arabic prepositional verbs like those given in (1). (In the transcription system here, I use the International Phonetic Alphabet (IPA); I only provide the meaning of the verbs because the prepositions are semantically vacuous.)

    (1) a. juɣi:r ʕala ‘invade, attack’
    b. janqadʕdʕ ʕala ‘attack’
    c. jaʕθur ʕala ‘find’
    d. jaqdʕi: ʕala ‘kill, finish, destroy’
    e. jaqbidʕ ʕala ‘capture, arrest’
    f. jutʕi:ħ bi ‘overthrow’
    g. jafʕal bi ‘act upon, affect, have an effect on’

The above verbs are seemingly intransitive since they must be followed by prepositional phrases (PPs); however, such verbs are in fact transitive. In support of this argument, I offer an account of the transitivity feature of such verbs both syntactically and semantically by examining their structural configuration and thematic representation. I will show that such verbs not only project, but also require, an internal argument (embedded under a PP-complement) to which they assign Theme, and that the preposition is semantically vacuous but serves a grammatical function (Case licenser), namely that the Theme argument is assigned genitive Case, and therefore structurally licensed, by the preposition. There are several syntactic diagnostics that can be used to determine the transitivity feature of such prepositional verbs; such tests include checking the obligatoriness of the PP-complements of such verbs and their derivatives (cf. Carrier and Randall, 1992), considering the relative order of postverbal arguments and adjuncts (cf. Pollard and Sag, 1987), and considering the distribution of complement PPs as opposed to adjunct PPs (Pollard and Sag, 1987). I will show that such diagnostics, among many others, provide evidence that the prepositional verbs in (1) are transitive (namely that they syntactically project an internal argument). Moreover, I provide a unified treatment of transitive prepositional verbs in Arabic and of the criteria based on which such transitive verbs must be distinguished from verbs followed by adjunct PPs (e.g., yajlis ʕala ‘sit on’, yanaam ʕala ‘sleep on’).

1:30-2:30Poster Presentations

An Afro-Asiatic perspective on the definition of serial verb constructions
Daniel Ross (University of Illinois at Urbana-Champaign)

An Afro-Asiatic perspective on the definition of Serial Verb Constructions Several constructions, in a range of Afro-Asiatic languages, present a classification problem based on the traditional definition of Serial Verb Constructions (SVCs), in part due to the non-concatenative morphology of these languages. SVCs are traditionally defined (cf. Aikhenvald 2006, Haspelmath 2016) as a monoclausal series of verbs which have shared features for Tense-Aspect-Modality and polarity, and are not connected by overt linking elements. Therefore, we can immediately exclude instances of the overt-linker type found in (1) (or converbal complex predicates of Ethiopian Semitic and Cushitic), despite these constructions being functionally similar to SVCs and sometimes called SVCs based on looser definitions:

(1) qaʕdat wa-katbat (Palestinian Arabic: Gamliel & Mar’i 2015:54–55)
    sit.PERF.1SG and-write.PERF.1SG
    ‘I was writing’

Still, there are other cases that are not canonical SVCs but hard to exclude by traditional criteria, requiring us to re-examine the limits of our definition to delimit a coherent phenomenon.

Case 1: In a number of modern colloquial varieties of Arabic there are verb pairs without overt linkers. But it is difficult to determine whether they count as SVCs by definition. One source for such constructions is omitted wa ‘and’ from (1) resulting in IMPF-IMPF or PERF-PERF combinations. Another is verb-complement constructions with differing tense/aspect-marking: if we consider the inflection contranstive in (2), it cannot be an SVC. The variation in this network of related constructions has not yet been fully documented across Arabic dialects.

(2) ʔaʕad ji-ħki (Jordanian Arabic: Ouali & Al Bukhari 2016:174)
    sit.PERF.3MSG 3M-talk.IMPF.3MSG
    ‘He kept talking.’

Case 2: Ait Seghrouchen Berber (Morocco) features aorist chaining (Bentolila 1981:151–171; Belkadi 2013:144–146), which is SVC-like except that it is formed with (and limited to) a particular verbal inflection (compare to Arabic above). Some would disqualify such restricted constructions as SVCs (Aikhenvald 2006:45). See also Pullum (1990) on the English “quasi-SVC” go get which cannot be inflected: *He go(es) get(s).

Case 3: Egyptian/Coptic has a remarkably long history of constructions that appear to fit the SVC template, sometimes in auxiliary-like usage, where two similarly inflected verbs express various functional relationships (Reintges 1995; Grossman 2009). Double-inflection forms recur over time, as if Coptic wants to be serializing, despite limited usage and anomalous semantic types; perhaps these are more like South Asian Agreeing Verb Constructions (Hock & Ross 2016) with the form of SVCs but not (necessarily) the same structure.

In conclusion, data from Afro-Asiatic languages can contribute to our general understanding of SVCs because these languages test the limits of the definition in ways that other languages do not. Continued documentation especially of colloquial varieties and dialectal variation is needed. These constructions suggest that the traditional definition of SVCs relies too heavily on form over function and structure: it would be productive to establish a wider category including not only the examples from the case studies above but also the overtly-linked multiverb constructions, while then establishing a typology of subtypes including traditional SVCs.

A typology of North African writing systems
Eve Koller (University of Hawaii at Manoa)

According to Simpson, “written language is a form of language. As such, it deserves to be treated with the methods of modern, scientific linguistic study which have been increasing our understanding of the spoken form of language for many decades” (Simpson 2001:11). Afroasiatic languages are unique in the long history and variety of writing systems which have been used to depict them. This poster presents the writing traditions of North Africa as a significant part of the Afroasiatic linguistic tradition, and as a valuable component of Afroasiatic historical linguistics.

The focus will be on the following writing traditions: Arabic, Egyptian hieroglyphs (Loprieno 1995), hieratic, demotic, Lybico-Berber (Pichler 2007), Phoenician-Punic (Krahmalkov 2001), and Tifinagh. As linguistic typological inquiry observes and attempts to describe the structural patterns of spoken language, in this poster I attempt to describe the patterns of the ancient orthographies of North Africa (as well as modern Tifinagh). These writing systems are compared to each other with regard to time period used, geographic distribution, type of writing system, number of symbols, distributions of symbols, and mediums of writing. In addition, the relationships between orthographies and their spoken languages are investigated. The intent of this poster is to invite further discussion of the linguistic typology of writing systems and linguistic descriptions of the interactions between language and orthography.

First language influence on the spoken Amharic of Gurage speakers
Solomon Getahun Assefa (Hawassa University)

This paper describes the form and use of Amharic by the Gurages. It investigates those areas in which Gurage is different and distant from Amharic and whether these differences and distance have any influence on spoken Amharic among Gurage speakers. Languages can be different from one another in almost every linguistic feature, like phonetics, phonology, phonotactics, syntax and semantics. There has been research in each of these features in various languages of Ethiopia in relation to Amharic. But as far as it has been noticed, no such research has ever been done on the linguistic relation between Gurage and Amharic at least not in the case of what and how linguistic features of Gurage languages influence the spoken Amharic among Gurage People. To this effect, the paper attempts to provide answers to the following basic questions:

  1. 1. What are the specific phonological features that characterize the Amharic spoken in Gurage?
  2. 2. Which morphological and syntactic features are characterized the Amharic spoken in Gurage?
  3. 3. What are the semantic, pragmatic and lexical peculiarities of the Amharic spoken in Gurage?

Since it is a very limited level of investigation with short duration, it has not been possible to do any experimental research or laboratory work with large numbers of Gurage peoples. Rather an interview method has been adopted and 24 Gurage peoples have been interviewed for 30 minutes to one hour. The interview situation has not been controlled and in the conversations various issues like culture, education etc. issues have been discussed in Amharic. All the interviews have been well recorded for the purpose of analysis. All the informants are ethnically Gurages and residents of Gurage Zone.

Thus, the corpus data comprises a collection of spoken texts gathered from informants who live in Gunchire1 town of Gurage Zone. The spoken Amharic data gathered through interview was segmented, transcribed, and annotated using tools like AUDACITY and ELAN, and analyzed thematically.

This paper serves as a foundational research for the creation of new speech variety of Amharic. It is also hoped that it has significant for research in language contact and change, as well as for a more detailed description of Amharic. What is more, it will add some knowledge in the applied aspect. Put differently, the findings of this study could be of assistance in understanding the performance of speakers/learners in second language learning situation and that would assist applied linguists in curriculum design or other pedagogical purposes. And again, it serves as a reference in studying other varieties of Amharic with similar linguistic and cultural situations.

This paper is restricted to the Amharic variety spoken in the Gurage Zone. Its scope is on three sample Gurage towns: Welkite, Buttajira, and Gunchire. Other Gurage towns, if considered, are discussed in relation to the three sample Gurage towns. The Amharic variety spoken in Addis Ababa2 city is considered to represent the so called standard Amharic for comparison.

1 Gunchire town is one of my research sites in Gurage Zone, Southern Ethiopia.
2 Zelealem (2007: 1) suggests that Addis Ababa Amharic is a standard Amharic.

Intensive word borrowing between Highland East Cushitic (HEC) and the Ometo group: Implications for the spread of concepts and influences
Shiferaw Assefa

The interactions between Highland East Cushitic and Ometo communities were complex and extended over many centuries, and the history of these interactions is reflected in the different kinds of word borrowing have taken place between HEC and Ometo speakers as a result of these contacts. The Highland East Cushitic speech communities, the Sidama, Kembatta, Hadiya, Burji, and Gedeo, currently reside along the Rift Valley south of the Awash River, with their lands intersecting those of Ometo peoples and adjoining to the north those of the South Ethiosemitic group.

As it can be indicated by loanwords, the Highland East Cushitic societies came under strong influence from Omotic communities who expanded into the lands of Highland East Cushitic people. The Omotic loanwords borrowed into Highland East Cushitic core vocabulary clearly show the intensity of intrusion and the influence. Similarly intense influences flowed the other direction as well, indicative of the assimilation of many formerly Highland East Cushitic-speaking people into the proto-Ometo society. There was also a similar impact of Ometo communities on the Burji who moved to the south, although it is not clear how early this encounter took place.

One interesting semantic difference in the early contact periods is the borrowing of the core vocabulary words for ‘heart’ and ‘liver’ from the Highland East Cushitic into Proto-Ometo, but for ‘moon’ and ‘stars’ from Proto-Ometo into Proto-Highland East Cushitic (PHEC). A question for future research is whether these patterns might reflect the spread of concepts about being and existence one way, from PHEC to Proto-Ometo, and another set of ideas, perhaps having to do with celestial imagery of the spirit realm, going the other way, from Proto-Ometo to PHEC.

Submorphemes in Hebrew
Zev Bar-Lev (San Diego State University)

Submorphemes in Hebrew
An innovative general theory of “submorphemes” in Hebrew (and beyond) is here proposed, to account for phonesthemes etc.

Words to Submorphemes.
The “morpheme” (now universally accepted) was itself Bloomfield’s 1926 innovation, preserving the Arbitrariness that de Saussure claimed for words, by focusing in on this smaller unit. Later, others posited even smaller units, submorphemes (e.g. phonesthemes). And nanosyntax (see Michal Starke’s on-line “Primer”) would imply the existence of submorphemes by claiming that simplex lexical items like Die can be (nano-syntactically) complex, occupying whole subtrees.

Phonesthemes are lexical clusters, like GL in GListen, GLeam, etc. Articles on them have broken through into mainstream linguistics journals, but without answering basic questions:

  1. 1. Are phonesthemes always initial two-consonant clusters, or can they have other constituents?
  2. 2. Are they onomatopoetic, or even meaningful? Assigning individual meanings to phonesthemes is extremely challenging.
  3. 3. Are they universal, with or without meanings?

Although consonant clusters arise in Hebrew as a result of vowel-drop (how can they be lexical clusters?), , Hebrew will be our principal language of investigation, because: 1. bi-consonantal root theory is a heterodox subfield of Hebrew linguistics, parallel to phonesthemes. 2. Hebrew does exhibit the striking phonesthemes like DV in DVorah ‘bee’ and DVash ‘honey’ (obviously lexical, and onomatopoetically, Dv- is a good imitation of bees buzzing.).

Our claims will include:

  1. 1. Observation of contrasting functionality distinguishing consonants from vowels, in Hebrew and beyond;
  2. 2. Observation of vowel motility even beyond Semitic. Among the other insights crossing the divide of language families is the linguistic explanation of “why Gold is expected to Glisten,” noting a linguistic parallel between GoLd and GListen.

On the universality of submorphemes, we will exemplify many lexical parallels, “hiding in plain sight” within and between languages.

Universal Meanings.
On the most ambitious question of universal meanings, our most challenging example will be GL, broadly recognized by researchers as meaning ‘bright light.’ Obviously this meaning does not fit Hebrew examples like GLidah ‘ice-cream,’ GLol ‘roll’), but Arabic aGaL ‘quick. exemplifies the pervasive semantic communalities found.

The proposed theory of submorphemes “modifies” de Saussure’s “axioms” little more than various branches of post-Chomskyan and even pre-Chomskyan linguistics. Its innovations like vowel motility lead to new implications in typology and lexical semantics. One indirect example of the way in which Semitic and non-Semitic language structures are brought together is seen in the stenographic system called “Speedwriting,” (spdwrtng) in which vowels are omitted as in Semitic spelling. The comprehensibility of speedwriting in non-Semitic languages supports using Hebrew as starting point for new insights that apply to Hebrew and beyond. For Saussure, He. even could have meant ‘tree.’ Is this really so?

Morphosyntactic evolution in the Moroccan Arabic verb
Thomas Leddy-Cecere (UT Austin) and Michael Turner (UT Austin)

In this investigation, we identify a potential case of inflectional debonding in Moroccan Arabic, whereby a previously bound verb stem and inflectional prefix complex undergo reanalysis and attain free status as morphosyntactically discrete elements.

This proposal is prompted by an emerging minority convention in Moroccan Arabic orthography. Consistent with accepted morphological analyses (Harrell, 1962), most speakers represent the imperfective verb as a single word:

1)  <kaytmana> / <كيتمنى>
    ka- j- tmənna
    ‘he wishes’

However, many instead represent the verb as two distinct units, placing a word boundary between prefix complex and verb stem:

2)  <kay tmana> / <كي تمنى>
    ka -j tmənna
    DUR-3M wish
    ‘he wishes’

That this novel orthography reflects a linguistically demonstrable reanalysis for these speakers is supported both phonologically and syntactically. First, we present evidence that it is not a straightforward syllabic rendering and may affect the application of word-internal assimilation rules; moreover, the same juncture is a common locus of pause in oral production, typologically atypical of bound morpheme boundaries (Whaley, 1997). Additionally, we identify cases in which a single prefix complex is utilized to govern two distinct, uninflected verb stems in a coordinated structure, strongly arguing for the free syntactic status of the elements involved.

We propose this reanalysis to be facilitated by historical sound change affecting Moroccan Arabic (cf. Versteegh, 2001), resulting in homophony between the (free) perfective and (bound) imperfective stems in most underived verb classes. This homophony allows for coidentification of the two as a single, unbound element utilized in both aspectual contexts and consequently separable from the prefix complex. As such, this development provides a possible case study of phonologically-motivated shift in morphosyntactic type, a topic of particular relevance to Afro-Asiaticists working to diachronically reconcile the diverse morphological schemes attested by the phylum.

Internal structure of root meaning in Afroasiatic languages
Karim Achab (University of Ottawa)

Bohas & Dat (2007, B&D henceforth) propose an organization of the lexicon based on a feature matrix and an etymon. The etymon is composed of two unordered phonemes expressing the feature matrix and its related notional invariant (NI henceforth). In the present study, adopting B&D’s NI, we investigate the internal structure of root meaning in Afroasiatic and propose that the root structure parallels the standard syntactic tree with a head, a specifier and a complement (see also Lahrouchi 2010). Unlike B&D, we take NIs to match discrete features borne by discrete phonemes, rather than scattered along a matrix of features borne by two distinct phonemes (etymon).

We adopt a comparative and constructional approach of Afroasiatic investigating a lists of words referring to body parts. For each root, we isolate the NIs associated with the phonemes and their phonological features. For instance, words indicating ‘nose’ such as anzar (Amazigh), anf (Semitic), naḫür (Biblical Hebrew) and hanci (Hausa) share the [nasal] feature and the phoneme [n] which express the NI ‘nasality’1, but the rest of the phonemes are different. For instance, are the words anzar and anf etymologically related or were the phonemes /z/ and /f/ chosen because they express the same NIs? The Amazigh noun anzar ‘nose’ contrasts with ansar ‘blowing one’s nose’. Many roots in Semitic and Amazigh show that the phonemes [f] and [s], respectively, are associated with ‘blowing/exhaling/flowing’. Accordingly, ‘blowing’ is the NI associated with the feature [strident] within these phonemes2. The words anzar and anf also contrast with respect to the sound /r/. The word anzar has morphological cognates in Semitic (Biblical Hebrew naḫür ‘nose’ and Arabic ḫinzir ‘pig’)3. The segment /r/ is found in words indicating ‘river’, which also contains or specifies the idea of ‘flowing/blowing’. We conclude that in words such anzar the segment /n/ indicates a locus, /z/ an event and /r/ a manner. Moreover, the syntactic tree proposed for the root involves segment hierarchy, i.e. which segment is the head and which ones are satellites. The positions order in the tree suggests that /s/ is the head, /r/ its complement and /n/ its specifier. These facts lend support to the hypothesis that /s/, as events, is a heads; /n/, indicating locus, is a specifier, while /r/, a manner component is a complement. Finally, the structure proposed offers a new way to look at the root extension theory suggested in the literature (Jagger 1988, Frajzyngier 2002; Militarev 2003 and Ehret 2008) among many others.

1 Also English ‘nose’ and French ‘nez’.
2 B&D (2007:116-117) attribute this NI to the segment /f/ in Semitic.
3 This suggests that ḫinzir originally meant ‘pig’s snout’ before it was extended to ‘pig’ by metonymy. Likewise, the Semitic word anf also has a cognate in Amazigh (ḫinfis ‘hedgehog’), which suggests that ḫinfis also originally meant ‘hedgehog’s snout’, extended to the animal by metonymy. Accordingly, the roots NF and NZR/NḪR existed in both Semitic and Amazigh but specialized and used differently.

4:30 Keynote Address:
Expanding the Afroasiatic Reconstruction: Historical Linguistics in a Deep-Time Language Family
Christopher Ehret (UCLA)

Sunday, June 3

Tense and aspect in Hamar
Binyam Sisay Mendisu (UNESCO-IICBA)

This study deals with the linguistic description of the tense and aspect system of Hamar. Hamar is an Aroid language of the Omotic language family and it is an SOV language. It is spoken by close to 46,000 native speakers. The people are mainly semi-pastoralists and the majority live in the South Omo Zone of the Southern Nations Nationalities Peoples regional state in Ethiopia. The paper examines the different ways in which tense and aspect are expressed in the language. The study also includes discussion of the Hamar’s tense/aspect system in light of typological observations made by different authors including Bhat (1999) and Comrie (1985). It has been noted that the language has two sets of different systems in the affirmative and negative paradigms. In the affirmative Hamar has five tense/aspect categories, namely simple past, remote past, present perfect, present and progressive. On the other hand, in the negative, the tense/aspect categories are reduced to three giving an evidence of neutralization (cf. Binyam 2016). The way in which tense/aspect is marked in the language ranges from the use of morphemes to the use of complex verbal forms which include existential verbs.

Serial verb constructions in Sezo
Girma Mengistu Desta (Addis Ababa University)

The paper discusses the compositional and semantic characteristics of Serial Verb Constructions (henceforth SVC) in Sezo, a little-known Afroasiatic language belonging to the Mao subgroup of Omotic family.

Most often Sezo SVCs are formed by simple juxtaposition of two verbs without any marker of coordination or subordination. SVCs with three verbs have also been recorded. These however, are analyzed as binary because two of the three verbs form a compound occurring in one syntactic slot. The components of an SVC are always contiguous and do not allow any syntactic constituent to appear between them. An SVC forms one phonological and grammatical word which describes a single event. Subject, tense, polarity and other syntactic dependency markers occur only once per SVC.

Following compositional characteristics of SVCs established by Aikhenvald (2006), I describe SVCs of Sezo as asymmetrical because they consist of a ‘major’ and a ‘minor’ verb. The ‘major’ verb comes from semantically unrestricted class and acts as the head of the construction. The ‘minor’ verb, which usually comes from semantically restricted class (e.g. a motion or a posture verb) has different functions. It changes the valence of the major verb or adds directional or aspectual values.

Taking the syntactic position of the ‘minor’ verb as the formal criterion, two types of asymmetrical SVCs are distinguished: SVCs in which the ‘minor’ verb precedes the ‘major’ verb and SVCs in which the ‘minor’ verb follows the ‘major’ verb. In the first type of SVCs, the function of the ‘minor’ verb is to add aspectual value to the major verb. In the second type of SVCs, the function of the ‘minor’ verb is either to change the valence of the verb or to express directional orientation.

Verbal extension and valence in the Gumer variety of Gurage
Fekede Menuta (Hawassa University)

The aim of this study was to describe how verbal extensions affect the valence number of verbs focusing on the derivation of causative, passive/reflexive, applicative, reciprocal and stem formatives. The study followed qualitative cross-sectional design with following structuralism approach to linguistics description. Though most of the Gurage language varieties largely have similar morpho-syntactic operations, the present study was based on the Gumer variety for purpose of consistency in transcription. We used IPA symbols and phonemic transcription which was glossed with Leipzig's morpheme-by-morpheme approach. The data were elicited from six key informants from Arekit town in Gumer district. It was found that the causative {a-} and double causative {at-} increases argument by one and two, respectively. The morpheme {jə-}, which has dative and ablative meaning, increases an argument at the object position. The applicative locative, instrumental, benefactive and malefactive increase arguments of a verb. The locative and instrumental are marked with a homophonous morpheme {bə-}. The benefactive and malefactive are derived with {-n(l)-} and {-b(w-}, respectively. Valence decreasing operations are passive/reflexive extension {tə-} and the impersonal passive extension {-i-}. Frequentative/intensive and reciprocal extensions maintain the valence of a verb the same. It was found that there are stem forming extensions that may increase or decrease valence as in aʤəgərəm 'caused trouble', tə-ʤəgərəm ‘be in trouble'. The form *ʤəgərəm’ without the prefixes has no meaning. It implies that both the root and the prefix are bound. Copulative constructions are problematic to be considered verbal extension since they are formed from nouns or adjectives. They are also problem to be considered an argument thereby increasing, decreasing or maintaining valence because they carry verbal element.

The marked nominative in Arabic, Semitic, and Afroasiatic
Lutz Edzard (University of Oslo)

In standard descriptions of Semitic, the nominative has the role of the unmarked case, typically in the role of the subject, whereas the accusative constitutes the marked case, typically in the role of the direct object. This paper argues that the concept of a “marked nominative” language, which is highly relevant on the Afroasiatic level, also plays an important role in Semitic and especially Arabic, as it catches many roles of the “accusative” (dependent case) that have no connection to the role of the direct object.

Generally speaking, the term “marked nominative” refers to a scenario, in which the nominative constitutes a longer or more complex, or in other words “marked”, form vis-à-vis the accusative. In Gothic and Old Norse, for instance, the nominative exhibits more complexity as compared with the accusative (1):

(1) Nominative vs. accusative in Gothic and Old Norse
            Nominative 	Accusative  Gloss
Gothic      dag-s       dag         ‘day’
Old Norse   arm-r       arm         ‘arm’
A comparable scenario is found in Harar Oromo (cf. Owens 1985: 101, 251) (2):
(2) Nominative vs. accusative in Harar Oromo
a.  sárée-n         adii-       ni   iyyi-t-i
    dog-NOM         white-NOM   FOC  bark-F-IPF
     ‘The white dog is barking.’
b.  haat-tii     okkóttée       goot-t-i
    mother-NOM   pot.ACC        make- F-IPF
     ‘Mother is cooking (lit. making the pot).’

Next to Cushitic and Omotic, comparable systems also obtain in Nilo-Saharan languages, thus pointing to an areal feature (cf. Dimmendaal n.d.: 1). In the latter case, the nominative is marked by a different tonal pattern (3):

(3) vs Nominative. accusative in Eastern Sudanic (Nilo-Saharan)
a.  έ-dɔ́l       έmbártá
    3SG-see     horse.ACC (two rising accents)
    ‘he sees the horse’
b.  έ-dɔ́l       έmbartá
    3SG-see     horse.NOM (one rising accent)
    ‘he sees the horse’

The proper terminology to adequately describe the form and function of the ”accusative”, a term that only captures a small subset of the variety of functions of the Arabic naṣb, remains a problem (cf., e.g., Haspelmath (2009)). While the term “nominative“ appears to be relatively unproblematic, one can point out that the term “accusative” represents a semantically narrowed translation of the more appropriate Greek term αἰτιατική by the Roman polymath Varro. As we will see, marking the direct object is but one function of “accusative” or, better: the “dependent case”; other functions include the marking of predicates, focalized subjects, and even the citation form and the vocative. In short: the functions of the case traditionally called “accusative” by far transcend the marking of the direct object, and the term “absolutive” is therefore preferable, as this term in linguistics refers to the unmarked citation form in “nominative-absolutive languages”. We will not touch upon ergative systems in this paper, as found, for instance, in modern Neo-Aramaic, even though these could be considered a further development of a “marked nominative” system.

Discourse particles in contemporary spoken Maltese
Lily Okalani Kahn (University College London) and Riitta-Liisa Valijarvi (University College London)

The purpose of our paper is to provide the first examination of discourse particles in contemporary spoken Maltese (e.g. mela ‘like, so, of course’, ifhem ‘you know’, heq ‘well’, ċoe ‘that is, like’, owkej ‘okay’, isma’ ‘listen’, speċi ‘like, kind of’, ux iwa ‘yeah’, em ‘um’, lajk ‘like’, well ‘well’, taf kif ‘you know?’). A discourse particle, also known as a pragmatic marker, is an element in speech used to indicate pragmatic functions, such as emphasis, attitude, to signal attentiveness and hesitation, and to smoothen and structure discourse (Schiffrin 1988, Fischer 2006).

While Maltese grammar has long been an area of interest for linguists (see Borg, Caruana, and Vella 2013, Saade and Tosco 2017 for recent examples) and there are two grammatical descriptions of Maltese (Borg and Azzopardi-Alexander 1997, Ambros 1998), our study will fill a gap in the research by producing a preliminary qualitative analysis focusing on an area that has not been explored in detail previously. We will classify the particles according to their discourse, their syntactic role and placement, their formal properties, and their historical origins (Siculo-Arabic, Sicilian and other varieties of Romance, English). For example, from Siculo-Arabic we have ifhem ‘you know’ and isma’ ‘listen’; from Romance we have ċoe ‘that is, like’ and speċi ‘like, kind of’; and from English we have lajk ‘like’, well ‘well’, and owkej ‘okay’. The data comes from native Maltese speaking consultants and is analysed following the principles of linguistic typology.

Our findings will fit into the existing research on discourse particles in cross-linguistic perspective (see e.g. Bayer and Struckmeier 2016, Fischer 2006), and will particularly complement research into discourse particles in other Semitic languages (see Alazzawie 2015 and Marmorstein 2016 for discourse particles in Arabic; see Maschler Dori-Hacohen 2012 and Gonen, Einat, Zohar Livnat, and Noam Amir 2015 for discourse particles in Hebrew). Maltese offers an unusual case study of discourse in a multilingual Semitic context due to its long history of language contact with Romance and English (cf. Vella 2013).

The following examples illustrate these points. In (1), speċi ‘like’ is used to smoothen and hesitate. In (2), isma' ‘listen’ is used to change topic, draw attention to the new topic, and smoothen the delivery of bad news. In (3), ifhem ‘well’ is used to convey a lukewarm attitude and imply that some contrasting information is to follow. In (4), taf kif is used to seek acknowledgement from the interlocutor. In (5), okay or mela serves to introduce a new sequence in the discourse.

(1) Taħseb li, speċi, terġgħu flimkien?
    Do you think that you'll, like, get back together?
(2) Isma', ma naħsibx li jirnexxieli niġi.
    Listen, I don't know if I'll be able to make it.
(3) Ifhem, ma kienx daqshekk tajjeb.
    Well, it wasn't that good.
(4) Kont, speċi, vera sorpriż meta smajt. Taf kif?
    It was like, you know, really surprising to hear that, you know?
(5) Okay/Mela, ejja nergħu nibdew mil-bidu.
    Okay, let's start over again.
Spatial deictics and grammaticalization in Tigrinya
Shimelis Mazengia Beyene (Addis Ababa University)

The concern of this paper is to examine the morphological structures, semantic features, syntactic functions and pragmatic uses of spatial deictics as well as grammaticalization of the definite article in Tigrinya—a member of the northern group of Ethiopian Semitic languages. The Tigrinya deictic expressions which involve encoding demonstratives are contrastive—proximal and distal. They have pronominal, adnominal and adverbial functions. As pronouns and noun modifiers, they inflect for gender, number and case while as locational and manner adverbs for case. In addition, pronominal and adverbial deictics may take enclitics. A deictic noun modifier, with either a proximal or distal status and inflected for gender, number as well as case, may be interpreted as an adjective or a definite article. If, however, it is repeated after the noun, it is interpreted only as an adjective. A deictic element in a relative clause has the function of a definite article. The fact that the deictic expressions which function as the variant forms of the definite article are those of the demonstrative counterparts shows that grammaticalization of the definite article is underway. The Tigrinya deictics which share the same roots of the proximal and distal forms are phonologically unbound but variable with the roots affixed and possibly encliticized.

Linguistic distance and mutual intelligibility among South Ethiosemitic languages: A combined approach
Tekabe Legesse Feleke (University of Verona)

Ethiosemitic languages are variants of the Semitic language family which are spoken in Ethiopia and in Eritrea. They are classified into North and South Ethiosemitic. The North branch consists of Ge'ez, Tigre and Tigrigna while the South Ethiosemtic includes Amharic, Argoba, Harari and several Gurage varieties. Many of the South Ethiosemtic languages are closely related, and the speakers of one variety can sometimes communicate with the speakers of other varieties. The relative distance (Bender, 1971; Hudson, 2013) and mutual intelligibility (Ahland, 2003; Gutt 1980) among these languages received some attentions. There are also proposals of the classification of the languages based on shared morphological and grammatical features (e.g. Demeke, 2001; Hetzron, 1972, 1977; Laslau, 1969). However, the previous studies have several shortcomings. First, they are limited in terms of scope-especially the Gurage varieties received a marginal attention. Second, some of the studies are not supported by sufficient data (Demeke, 2001; Hetzron, 1972). There are also inconsistencies among mutual intelligibility studies (cf. Ahland, 2003 and Gutt, 1980) which possibly emanate from the inherent methodological limitations and complex intermingling among the languages.

The present study employed a combination of measures: structural distance (lexicostatistics and Levenshtien distance), mutual intelligibility, perceived distance (judged similarity and judged intelligibility) and geographical distance to determine the distance and mutual intelligibility among 10 South Ethiosemitic languages: Chaha, Ezha, Harari, Silt’e, Gora, Gumer Soddo, Muher, Ennomer and Mesqan. The study intended to (1) determine the relationship among the distance measures and (2) re-examine the previous classification of the languages. The lexical and Levenshtien distances were computed based on words in the translations of a fable-the North Wind and the Sun. The lexical distance was determined by computing the average of the percentage of non-cognate words within pairs of languages. The Levenshtein distance was determined by computing the cost- insertion, substitution and deletion required to transform the pronunciation of one cognate to another, using GabMap (see Nerbonne et al., 2011). Word categorization test was used for mutual intelligibility measure (Tang and Heuven, 2009). The test requires categorizing words under their Semantic categories; for example apple under FRUIT. The perceived distances were determined using the recordings of the translations of the North Wind and the Sun. After listening to each recording, selected participants determined (a) to what extent each recording is similar with their native language (judged similarity) and (b) to what extent they understand each of the recording (judged intelligibility). GabMap was employed for the clustering and cluster validation (multidimensional scaling and bootstrapping). The geographical distance among the language areas was obtained from Google Earth using GabMap.

The regression results show strong correlations among the distance measures (structural, mutual intelligibility and perceived distance). The clusters obtained from the measures are also largely similar. These results are consistent with previous correlation reports (Gooskens, 2013; Gooskens and Heeringa, 2004; Tang and Heuven, 2009). The clusters obtained from the measures are also very similar with the genetic classifications proposed by Demeke (2001) and Hetzron (1972). However, Harari and Soddo have shown deviations from the previous classifications. This divergence is associated with the influence of Cushitic languages. Strong association was also found between language distance and geographical distance which indicates the influence of language contact.

Language contact and its effects on Gurage varieties of Muher
Awlachew Shumneka Nurga (Addis Ababa University)

Social and linguistic contacts between speakers of diverse varieties as well as the influence of the surrounding Cushitic languages contributed to the establishment of Gurage dialects and widespread bi- or multilingualism. However, the actual extent and the effects of language contact on individual languages in the Gurage Zone are not clearly known. Therefore, the main objective of this research is to investigate language contact and its effects on language use and form on Gurage varieties of Muher.

The Muher community lives in the north-western part of the Gurage Zone. Its neighbors are Ezha in the west, Mesqan and Dobbi in the southeast, Wolane in the northeast, K’abeena in the northwest and Silt’e in the southeast. As a result, many of the Muher speakers are bilingual in one of these languages. The study targets the language behavior of individuals belonging to Muher Gurage ethno-linguistic group residing in rural and urban settings, namely the rural areas of Teklehaimanot, Zəbbidar and the town of Hawarijat and Wolkite as zonal administration center. Mixed research methods (questionnaire, interview, and participant observation) are used as research tools.