Universität Potsdam

Formal Semantics of African Languages

Modularbeit SoSe 2022
Anna Struck

Jula Comparative Constructions

In the English grammar the comparative in phrases like “Mary is taller than John” or “Susan is
more intelligent than Bob” is built by adjoining a phrase headed by the comparative element
“more” to the part of the phrase it is compared to or by introducing a than-clause with its
object. This is a common procedure for many European languages. By using a comparative
(COMP) particle or word that connects the compared object and the comparee. West-African
languages like Yoruba, Akan, Jula, Hausa or Ga (just to name a few) employ a very different
strategy to express the comparative. Cross-linguistically the major strategy for expressing
comparison in West-African languages is using a free-standing lexical item with a meaning of
“surpass” or “exceed”. (Stassen, 1985) These are exceed-type languages, assuming an exceed-
relation that connects two degrees to each other.

This paper presents an overview over comparative constructions in Jula and their variations.
Jula is an exceed-type language, building the comparative with a verb connecting to NPs with
one surpassing/ exceeding the other in a specific area or manner. This strategy of expressing
the comparative is common in many African languages. The translation of the verb that
constitutes the comparative in the examples in this paper is always “surpass”, possibly due to
the influence from the French word “surpasser”. This type of degree sensitivity was described
in Beck et al (2009). In the paper they describe three parameters for classifying languages
according to the type of comparative construction that is used. Exceed-type languages have
hitherto been classified to have [+DSP][-DAP] setting. In the following section I will show that
that is not the case and Jula has at the least a [+DAP] feature. All examples were elicited from
a native speaker of Jula over three sessions, one of which was a final check of the material.
Preliminaries

Jula (Dyula) is a language that can be traced back to the Mali empire (now Guinea and Mali)
and spread with the traders looking for gold. Dyula literally translates to “trader” and today it
is spoken by roughly 300 000 people. Dyula is a dialect of the Mande language, spoken widely
over much of West Africa and the Ivory Coast. Due to a history of travelling and the closeness
to other West-African languages, the Dyula people can easily understand many other dialects.
The grammar is simple, employing a TAM system for verbs and no gender or noun classes, but
the information conveyed in discourse and context is thus far more complex. Dyula also has a
rich vocabulary with loan words and influences from French and Arabic.

A common feature for many African languages is the lack or very minimal use of adjectives.
Especially West-African languages like Akan and Creoles like Krio use verbs to describe
attributes or states. In Krio the predicate “is red” from “The dress is red” is a single word, a
verb stating a state of “being red”. Jula similarly does not have many adjectives. The few
adjectives it does have are possibly due to the influences from French or Yoruba. So in order
to describe a state, an attribute or a feature, a nominalization suffix ”-ya” is appended to the
verb. See (1) for an example.

(1) Fofana jan-ya be tɛmɛ 1.80 m kan.

Fofana tall-NOM HAB surpass 1.80m PostP.
‘Fofana is taller than 1.80m’

Another feature common to many African languages is context dependency. A smaller

vocabulary does not imply simplicity. It just means that so much more information comes from
context and leaves room for several interpretations and multiple meanings of words in
different constructions. In this case most adverbs and adjectives in Jula can be omitted if the
context of the phrase is clear. Compare (2a-b), context makes it obvious that the comparison
is done on a scale of speed in (a) and on a scale of distance in (b). Thus there is no need to add
the adverbs for “faster” or “further” in order to understand the meaning of the phrase.
(2)
a. Fofana boli be 2m/h tɛmɛ.
Fofana running HAB 2km/h surpass.
‘Fofana runs faster than 2km/h.’
b. Fofana boli be 2km tɛmɛ
Fofana running HAB 2km surpass
‘Fofana runs more than 2km.’

Furthermore there are two copulas in Jula “ka” and “bɛ”. “Ka” is a predicative copula and “bɛ”
is an existential or locative copula used to express the notion of “having” something. See (3a-
b) for examples.

(3)
a. Wari bɛ yan
money COP here
‘There is money here.’

b. Wari bɛ Adama fɛ
money COP A. PostP
‘Adama has money’ (lit. Money is with Adama)

Both copulas, as well as all aspect markers, in Jula have negative counterparts. Where,
especially western languages employ a negation particle or affix to express the negated
version of a verb or a phrase, in Jula the copulas and aspect markers have negative
counterparts expressing a state of “not being” or “not having” something. The counterpart for
“ka” is “ma” (4a) and the counterpart for “bɛ” is “te” (4b).

(4)
a. Ka Hawa jati, Fofana ma jan.
INF Hawa consider, Fofana COPneg tall.
‘Compared to Hawa, Fofana is not tall’

b. Fofana jan-ya te 1.80m dan.

Fofana tall-NOM COPneg 1.80m surpass.
‘Fofana is not taller than 1.80m’
Comparative – an overview
There are three strategies for expressing the comparative in Jula. All are built with a verb
meaning “surpass” or “exceed”. The first construction in (5) combines an infinitive marker with
the verb for surpass followed by a preposition.

ka tɛmɛ __ kan
(5)
a. Fofana ka surun ka tɛmɛ yirimͻgͻnin kan
Fofana COP short INF surpass statues on
‘Fofana is shorter than the statues.’

“Kan” introduces the object that is compared. “Ka tɛmɛ” can appear in the construction
illustrated in (5) or it can appear sentence-finally without the preposition (5b - 6). However,
they cannot be used interchangeably. In (6) “kan” is omitted because the objects compared
(Kofis money-spending peoples money-earning) do not belong to the same conceptual group.
Using “kan” would compare Kofi to the people (same conceptual group) instead.

b. Musa ka teli, nga Audu ka teli kà tɛmɛ.

Musa COP speedy, but Audu POSS speedy INF surpass.
‘Musa is fast, but Audu is faster.’

(6) Lon kelen kᴐnᴐ Kofi be wari caman bᴐ ka tɛmɛ

Day one inside Kofi HAB money much spend INF surpass

caman mᴐgᴐ be min sᴐrᴐ kalo kelen kᴐnᴐ.

people many HAB REL earn month one inside.
‘Kofi spends more money in a day than most people earn in a month.’

Ni __ ye

The second comparative construction Jula makes use of is “ni ye”, where the compared object
is situated between “ni” and “ye”. “Ni” is a conjunction and “ye” a postparticle and this
combination can only be used with adjectives. Together with a verb “ni” is interpreted literally
as a conjunction. Compare (7a-b) for exemplification.
(7)
a. Hawa ká jan ni Fofana ye
Hawa COP tall with-CONJ Fofana PostP
‘Hawa is taller than Fofana.’

b. Gninan-w be boli ni jakumani ye

Mouse-PL HAB run CONJ cat PostP
‘The mice are running with the cat’

Be tɛmɛ __ kan
The third construction is built with a habitual imperfective marker “be” and the verb for
“surpass”. Combined with “tɛmɛ“ this construction is similar in structure to „ka tɛmɛ” and the
compared object is introduced by a sentence-final preposition (8a). Unlike “ka tɛmɛ” the
compared object can be placed between the habitual and the verb (8b-c).

(8)
a. Fofana janya be tɛmɛ 1.80 m kan.
Fofana tall-NOM HAB surpass 1.80m PostP.
‘Fofana is taller than 1.80m.

b. Fofana boli be 2km/h tɛmɛ.

Fofana running HAB 2km/h surpass.
‘Fofana runs faster than 2km/h.

c. Fofana jan-ya be 1.80 m dan.

Fofana tall-NOM HAB 1.80m surpass.
‘Fofana is taller than 1.80m’

Intermediary summary
There are different structures for the comparative in Jula: an infinitival verb phrase followed
by a postposition, a comparative particle followed by an NP and a postposition and a habitual
marker preceding a verb phrase. Bambara, which is mutually intelligible with Jula, employs
the same infinitival verb phrase construction to express the comparative. (Charles-Luce, 1986)

1. INFINITIVE MARKER + VERB + NP + POSTPOSITION

2. COMPARATIVE PARTICLE + NP + POSTPOSITION
3. HABITUAL MARKER + VERB + NP + POSTPOSITION
Phrases (5-8a) are examples of these. Depending on the construction the verb surpass can
take up to three arguments – the subject, the compared object and a gradable property. The
order also varies depending on the type of construction chosen. The commonly employed
constructions with “ka tɛmɛ/ ni ye” show a regular SVO word order. In (8c) the subject is
compared to a degree and yields not only a different comparative structure but also a different
word order, where a full predicate appears sentence-final in the SOV order.
4. HABITUAL MARKER + NP + VERB

In (8c) the degree compared to (1.80m) is the direct object followed by the full predicate
“dan”. This is possible because of the nominalization in Jula. “Jan” – tall is nominalized to
“height” and appears in the subject position together with Fofana, literally meaning “Fofanas
height…”. This is similar in (8b). A full predicate appearing sentence-finally thus forces the
nominalization of the property in the subject position.

(9) Gninan-w be boli (joonan) ka tɛmɛ jakumani kan

Mouse-PL HAB run (ADV) INF surpass cat PREP-on
‘The mice run faster than the cat.’

(10) Hawa ká jan kà tɛmɛ Fofana kan

Hawa COP tall INF surpass Fofana on-PREP
‘Hawa is taller than Fofana’

The subject can consist of an object as in the case of (9) and (10), it can consist of an NP that
contains the subject and a noun or a subject and a nominalized verb as in (8b-c). Like Yoruba,
Jula also has predicative verbs that incorporate a copula and an adjective.

From Yoruba:
(11) Ade ga jo Isaac lo.
Ade is_tall exceed Isaac go
‘Ade is taller than Isaac’ (Beck, 2009)

Antonyms are also built with the verb for “surpass”. Take (12-13) for example. These phrases
describe attributes that are on the other end of the scale tall/short big/small and the verb for
“surpass” expresses a comparison with the negative antonym. The direction of comparison is
unidirectional.
(12) Fofana ka surun ka tɛmɛ yirimͻgͻnin kan
Fofana COP short INF surpass statues on
‘Fofana is shorter than the statues.’

(13) Fofana ka dͻgͻ ka tɛmɛ mͻbili kan

Fofana COP small INF surpass car on
‘Fofana is smaller than the car.’

Factors that influence the choice of construction

Several factors influence which comparative construction is chosen. “Ni ye” cannot be
combined with a verb, lest it gets a literal interpretation of a conjunction (see 7b). Instead “ka/
be tɛmɛ” is chosen. There is no indication that “ka tɛmɛ” can be combined with degrees. Ka
tɛmɛ compares two objects and for a comparison on a scale “ni ye” or “be tɛmɛ/be dan” seems
to be preferred. Animacy also plays a role in the choice of comparative constructions.
According to a native speaker of Jula the construction “ka tɛmɛ” sounds unnatural when
comparing an animate with an inanimate object. For this comparison “ni ye” is preferred.

The semantics of exceed-type languages

A formalization of gradable adjectives has been done by von Stechow for English. This includes
adjectives like ‘tall’ and ’intelligent’, adjectives that are on a spectrum (or a scale). According
to von Stechow, these gradable adjectives are of type <d> and compare an object’s property
to a degree.
Gradable adjectives (type <d,>; von Stechow):
[[ tall ]] = λd: d ∈ Dd.λx: x ∈ De . Height(x) ≥ d
[[ intelligent ]] = λd: d ∈ Dd.λx: x ∈De . Intelligence(x) ≥ d
[[surpass]] = lx. lPÎ D<e<d,t>>. ly. max(ld. P(x)(d)) > max(ld. P(y)(d))

Similar formalizations have been proposed for languages like Yoruba and Hausa that also have
an exceed-strategy. The verb for “exceed” or “surpass” is taken to denote a 2-place relation
between degrees, where the maximum degree of an objects property exceeds the maximum
degree of another objects property. These degrees are derived from the maximized content
of the gradable object applied to their arguments. (Zimmermann, 2009) Take the phrase from
(8c):
Fofana jan-ya be 1.80 m dan.
Fofana tall-NOM HAB 1.80m surpass.
‘Fofana is taller than 1.80m’

The literal meaning of this phrase is “Fofanas height exceeds 1.80m.” The full predicate “dan”
takes two arguments, relating two degrees d and d’ to each other, where d’ exceeds d.

[[dan]] = λd. λd'. d‘ surpasses d

The suject NP consists of Fofana and height, where [[height]] takes two arguments: a person
x and the degree d which the height exceeds.

[[height]] = λx. λd. height(x) ³ d

Inserting Fofana for the x as the argument of height yields

[[fofanas height]] = λd. height (fofana) ³ d

This is true if and only if the maximum height of Fofana exceeds the degree of 1.80m.

= 1 iff max(λ.d height(fofana) ³ d) > 1.80

Types of comparatives in Jula:

The analysis of the elicitations yielded several different comparative constructions. A
comparison with a degree relates two degrees to each other. In (12) it relates the degree of
Fofanas height to the degree which it has to surpass for the sentence to be true. In (13) the
degree of the waters deepness is related to the degree which it surpasses. A difference
comparative takes three degree arguments – Hawas height, Fofanas height and the degree of
Fofanas height minus 5cm. Degree questions open up the possibility for a language to ask for
a specific degree. Subcomparatives give rise to a comparison on two different scales. These
types of comparatives are relevant for classifying and analyzing languages and their
comparative structures.
A comparison with a degree (CompDeg)
(12)
a. Fofana jan-ya te 1.80m dan.
Fofana tall-NOM COPneg 1.80m surpass.
‘Fofana is not taller than 1.80m’

b. Fofana jan-ya te tɛmɛ 1.80m kan

Fofana tall-NOM COPneg surpass 1.80m PostP.

(13)
a. Ji dun-ya be 2m dan.
Water deep-NOM HAB 2m surpass.
‘The water is deeper than 2m.’

b. Ji dun-ya be tɛmɛ 2m kan.

Water deep-NOM HAB surpass 2m PostP.

Difference comparatives (DiffC)

(14) 5cm le be Hawa surun-ya ni Fofana ye.
5cm FOC HAB Hawa shorten-NOM CONJ Fofana PostP
‘Hawa is 5cm shorter than Fofana.’
Literally: It is 5cm that Hawa shortens compared to Fofana

Degree questions (DegQ)

(15) Kofi hakilimaya bɛ (cogo) di?
Kofi intelligence COP (way) how
‘How smart is Kofi?’

(16) Minabilaminan bon-ya bɛ (cogo) di ?

Closet big-NOM(being) COP (way) how
‘How big is the closet?’

Subcomparatives (SubC)

(17)
a. Minabilaminan ka jan ka tɛmɛ da lᴐsurun kan.
Shelf COP high INF surpass door width PostP.
‘The shelf is higher than the door is wide.’
Literally: ‘The shelf is high surpassing the width of the door.’

b. Minabilaminan jan-ya be da lᴐsurun tɛmɛ.

Shelf height-NOM HAB door width surpass.
‘The height of the shelf exceeds the width of the door.’
Semantic parameters

Becks et al describe three parameters for determining different comparatives and cluster
them into categories. A DSP-parameter concerned with lexical variation and a DAP-
parameter concerned with the syntax/semantics interface. A third parameter DegPP is
concerned with degree modifiers. (Beck, 2008) A definition for the parameters comes from
Beck and is given below:

Degree abstraction parameter (DAP): A languages ability to have a binding of degree

variables in the syntax
Degree Semantics Parameter (DSP): A language {does/does not} have gradable predicates
(type <d,> and related), i.e. lexical items that introduce degree arguments.
Degree semantics on nominal parameters – see comparison with a degree and difference
comparison.
Degree Phrase Parameter (DegPP): A language {does/ does not} have overt degree phrases.
The degree argument position of a gradable predicate {may/ may not} be overtly filled.
This parameter arises only for languages with a +DAP setting, because this position is filled
by expressions that trigger a binding of the degree argument.

These parameters have been used to classify languages that do not give rise to the usual
comparative strategies. Exceed-type languages have until now been classified as having the
[+DSP][-DAP] setting. These diagnostics do not apply to Jula. That Jula has the +DSP setting is
illustrated with phrases like (8). A language with a -DAP setting does not have degree
question or subcomparatives (Beck, 2008). This clearly is not the case for Jula, which shows
both.

Summary
The first section of this paper gave a short introduction to general features and structures in
Jula. Examples illustrate that the comparative can be built with three different structures
and what their variations are. Different types of comparatives like the subcomparative,
degree questions, difference comparatives and comparative with a degree can be found in
the language. A different approach is necessary to formalize languages with an exceed-type
strategy. Beck et al (2009) assume that “exceed” indeed has the same features as a
comparative particle [-er] and can therefore be analyzed as such.

These semantic macro-parameters (Beck et al, 2009) are proposed as a classification of

languages that employ different strategies for expressing the comparative as opposed to the
majority of European languages. The first two parameters are semantic, while the latter is
only semantic in nature, but governs the occurrence of degree phrases in overt syntax.
(Zimmermann, 2009) Any language that has comparison constructions is a [+DSP] language.
They are [-DSP] if the language does not make any reference to degrees in their grammar
and makes use of conjunction strategies to refer to a scale.1

Languages with a positive DAP-setting allow for subcomparatives and give rise to Negative
Island effects (NIE). These languages allow the formation of sets of degrees through
predicate abstraction over degree variables. (Zimmermann, 2009) Without a binding of
degree variables there would not be a possibility for degree operators and normal scope
operations. Degree operators that take wide scope need a binding of degree variables.
Measure phrases and negative island effects could not be proven with the elicitations for
this paper and it would be interesting to test for this in the future. More study could also go
the question, what other restrictions the comparative constructions underlie. “Ka tɛmɛ” is a
comparative that seems unnatural for comparing an object to a degree. “Ni ye” can only be
used with predicative adjectives, otherwise it is interpreted as a conjunction and not a
comparative. The habitual marker combined with the verb phrase appears to be the
comparative that is used when comparing with a degree or comparative construction that
classify as proof for the positive setting of the DAP parameter, like the subcomparative.

Exceed-type languages have been classified to have [+DSP][-DAP] setting. I wanted to show
that that is not the case and Jula has at the least a [+DAP] feature. Beck et al argue that a
language that has the negative setting of the DAP parameter will not have degree questions,
measure phrases and subcomparatives. This clearly is not the case for Jula, which displays
degree questions, subcomparatives and comparatives to degrees and therefore following

1
Conjunctive comparative strategies like in Motu employ a strategy where conjunctions like ‘and’ or ‘but’
connect the two NPs. For more information on this see Beck (2009).
the classification of the semantic parameters has a positive DAP setting. Further indicators of
these parameters which could not be proven from these elicitation are Negative Island
Effects (NIE) and Measure Phrases (MP). It would be interesting to look into this in further
studies.
References

Beck, S. (2008). Comparison Constructions. To appear in P. Portner, C. Maienborn, K. von

Heusinger (eds.), HSK Semantics. Berlin: Mouton de Gruyter.

Beck, S. et al. (2008). Crosslinguistic Variation in Comparison Constructions. Ms. Universität

Tübingen.

Beermann, D. Brindle, J. Hellan, L. et al. (2005). A Comparison of Comparatives. Proceedings

of the LFG05 Conference. CSLI Publications. University of Bergen. http:/csli-
publications.stanford.edu/

Bhatt, R. & Takahashi (2007). Direct Comparisons: Resurrecting the Direct Analysis of Phrasal
Comparatives. Proceedings of SALT 17. Cornell: CSC Publications. 19-36.

Charles-Luce, Jan (1986). Comparison in Bambar: An Infinitival Verb Phrase. Studies in

African Linguistics. Volume 17, 2nd Edition. University of Indiana

Stechow, A. von (2005). Different Approaches to the Semantics of Comparison. Ms.

University of Tübingen.

Zimmermann, M. (2009) The Semantics of Comparative Constructions in Hausa. Presentation

Eberhard-Karls University of Tübingen 23.01.09
What we do not find:
Negative Island Effects (NIE)
*Mary bought a more expensive book than no boy did.

MP: Measure phrases CHECK IF SIMILAR IN YORUBA

N is exactly 1.70m tall