1. Introduction

The literature on language change provides an abundance of research which demonstrates various tendencies and directions of development. In recent years, an increasing number of studies within the framework of grammaticalization and later on construction grammar, have shed light on the ways in which new linguistic expressions emerge and integrate into the lexicon and the grammar of language (Trousdale 2014, Traugott & Trousdale 2013, Hilpert 2013). Researchers working in construction grammar have modified their perspective of the way change is defined. Rather than considering the emergence of a new string of language as a case instantiating a discrete process of grammaticalization or lexicalization, they have started to endorse a position of a gradient output located on a continuum which ranges from grammatical or procedural constructionalization to lexical constructionalization. The end results of these processes may be associated with characteristics of both, thereby rendering the distinction between them less rigid than was previously assumed.

In accordance with this perspective, we find an interesting development of the Hebrew adjectives tov and French bon (good), which originally denoted a favorable quality and today exhibit polysemy, as illustrated in (1)‑(2):

(1)	a.	Hu	amar	she‑ha‑seret	haya	tov.
		He	say.pst.3sg.m	cpl‑def‑movie	be.pst.3sg.m	good
		‘He said that the movie was good.’

	b.	Kibalti	be‑’avoda‑ti	ha‑kodemet	maskoret	tov‑a.
		Receive. pst.1sg	in‑work‑1sg.pos	def‑previous	salary	good‑sg.f
		‘I received a good salary in my previous work.’
	c.	Bizbazti	axshav	kama	dak‑ot	tov‑ot	kedey
		spend. pst.1sg	now	a.few	minute‑pl.f	good‑pl.f	to
		lehaxin	et	ze.
		make.inf	acc	it
		‘I now wasted a good few minutes in order to prepare it.’
	d.	Tov,	namshix	la‑nose	ha‑ba?
		good	continue.fut.1pl	to.def‑subject	def‑next
		‘Ok shall we continue with the next subject?’
(2)	a.	Même une bonne éducation, autrefois garante d'emploi, aide rarement à l'heure actuelle.
		‘Even a good education, previously ensuring employment, rarely helps these days.’
	b.	Je gagne un bon salaire, mon travail est à la fine pointe de la technologie.
		‘I earn a good salary my work is at the pinnacle of technology.’
	c.	Au bout d'une bonne demi‑heure de recherche, on finit dans un resto comme tous les autres, avec un menu touristique !
		‘After a good half an hour of research, we finished at a restaurant like all the others with a tourist menu.’
	d.	Bon, je vais m'abstenir de faire des commentaires politiques.
		‘Okay, I will abstain from making political commentaries.’

In examples (1)‑(2), the polysemy of the adjectives is explained as follows: The adjectives in (a) sentences refer to a quality in the sense of something being favorable, good or valuable;1 in (b) sentences they convey a quantitative meaning; in (c) sentences they function as intensifiers which express an extension of the frame beyond that indicated,2 and in (d) they function as discourse markers signaling a variety of functions, such as acceptance, approval and topical shifts.

We also find other adjectives commonly known to be synonymous with tov and bon. These adjectives denote more specified favorable qualities which were originally associated with benevolence, valued character or appearance, such as Hebrew yafe ‘beautiful’, hagun ‘decent’, mexubad ‘honorable’ na’e ‘nice’, nadiv ‘generous’ and ra’ui ‘deserving’ and French belle, beau ‘beautiful’, digne ‘worthy’, décent ‘decent’, honnête ‘honest’ and honorable ‘respectable’ among others. These expressions display meanings which are similar to tov and bon in the sense that they have also acquired a metaphorical/broadened meaning of upscaling quantification, as in the (b) sentences, especially in contexts of salaries and benefits. However, they have not further acquired the functions of intensification and discourse marking, as in (c) and (d).

It should be stated that other adjectives denoting highly favorable qualities, such as wonderful, fantastic, remarkable and marvelous and their Hebrew counterparts have been excluded from this research. The reason for this exclusion is their inherent amplifying feature found already in their very first occurrences, which we assume precluded development of further intensification. Consequently, due to their extreme character they will more readily describe a large salary for example, without having to undergo a metaphorical or a broadening shift.3

The diachronic evolution of lexical items toward more pragmatic uses has been studied and documented extensively in French (Auchlin 1981; Beeching 2009; Brémond 2002; Dostie & Pusch 2007; Hansen 1995; Jayez 2004; Lefeuvre 2011; Morel & Danon‑Boileau 1998; Winther 1985 among others) and much less in Hebrew (Livnat & Yatziv 2003; Ziv 2006; Maschler 2009). The analysis advanced in this article seeks to focus on the comparative perspective and thereby to highlight some interesting parallel evolutions. Based on these preliminary observations, the purpose of this paper is as follows: 1. To trace the development of the meanings and functions of the Hebrew and French expressions mentioned above and to account for the difference in the evolution of the polysemy of bon and tov compared to the other adjectives. 2. To examine the extent to which the development of these adjectives has been motivated by an increase in the speaker’s subjectivity and intersubjectivity (Athanasiadou 2007, Traugott 2010, Narrog 2017). 3. To suggest that the final output of the development is perceived as convergence between grammaticalization and lexicalization processes.

In line with Construction Grammar, we suggest that the four uses of Hebrew tov and French bon can be summarized as the following:

[DET bon N] ↔ quality (un bon livre, sefer tov ‘a good book’)

[DET bon N (quantifiable)] ↔ quantity (un bon salaire, saxar tov ‘a good salary’)

[DET (numeral/quantifier) bon N (measurement unit)] ↔ intensity (quelques/cinq bonnes heures, kama/xamesh sha’ot tovot ‘a good few/five hours’)

[Bon] ↔ discourse marker (Bon, allons‑y, tov, bo’u nelex, ‘ok, let’s go’)

According to this formulation, the four meanings of the construction [tov] and [bon] consist of one substantive element which is the adjective tov or bon and two schematic elements, namely the determiner and the noun. We will show that the polysemy associated with these adjectives derives from a change in the two schematic elements according to the restrictions of the construction.4

2. Constructionalization and (inter)subjectivity

This section will now examine the two closely related types of diachronic processes, namely grammaticalization and lexicalization, and the more recent view of language asserting that the two in fact constitute a case of gradience.5 The section will further look at some current approaches to (inter)subjectification in order to comprehend the motivating force behind the change.

2.1 Lexicalization and grammaticalization

At the beginning of diachronic research, studies on grammaticalization spread like a wildfire (Givon 1971, Traugott 1978, Greenberg 1978, Harris 1979) and the distinction between grammaticalization and lexicalization was barely given focus at the time. Only twenty years later the literature on diachronic change started to differentiate between the two processes. Lehmann (1985, 2002), whose two articles demonstrate the evolution of the theory, is one of the precursors in pointing out the distinction.6 In lexicalization, research refers to the diachronic aspect of different word formations, such as compounding, blending, derivation, conversion, loan translation, back formation and coinage (Brinton & Traugott 2005). Grammaticalization is traditionally referred to as a process where in a certain context, a lexical expression tends to acquire a grammatical function and continues further to acquire functions as it develops (Hopper 1991, Hopper & Traugott 2003). This process is accompanied by several morpho‑syntactic characteristics, such as a transition from concrete referential lexical meaning to abstract procedural meaning as well as bleaching which refers to the loss of categorial meaning, layering which refers to the co‑existence of old and new meanings and persistence which refers to a reflection of the original lexical meaning in later grammatical functions.

Diachronic research which developed significantly within the last forty years resulted in an increased interest from a variety of approaches.7 Research within the framework of construction grammar suggesting that there is no clear divide between grammatical and lexical expressions has gained great focus. According to construction grammar, the basic building block of language is a form‑meaning pairing associated minimally with “semantics, pragmatics and discourse function on the meaning side, and syntax, morphology and phonology on the form side” (Traugott & Trousdale, 2014, p. 258) in varying degrees of size, shape, complexity and schematicity (Noel 2007, Goldberg 2011, 2013, Hoffmann & Trousdale 2013, Hilpert 2014 and Bat‑Zeev Shyldkrot 2017). The creation of a construction, i.e. the output of constructionalization, is “a new node in the language network that may be more towards the ‘contentful’ end of the continuum, or more towards the ‘procedural’ end” (Traugott & Trousdale, 2013, p. 149). Lexical constructionalization is associated mostly with major categories such as nouns, adjectives, verbs and some adverbs, whereas grammatical constructionalization yields procedural material which signals linguistic relations such as aspect, tense, articles and auxiliaries. Evidence suggests that a differentiation between grammatical and lexical expressions may be blurred when the output is partially lexical and partially grammatical.

2.2 (Inter)subjectivity

In exploring the motivation for the change in meaning and function of the expressions of the favorable qualities discussed in this article, special emphasis is given to the interaction between tendencies of (inter)subjectification and semantic change. The literature discusses different definitions of (inter)subjectification. Langacker (1990, 1991, 1998) developed the concept of subjectivity within the framework of cognitive Grammar. According to him, subjectivity is a matter of construal which focuses on an event and participants with relation to the ground and as such is more of a synchronic concept rather than a diachronic one (Narrog, 2012, p. 72).

A different approach is proposed by Nuyts (2001), who developed Lyons’s initial theory (1977). According to Nuyts, evidentiality refers to a subjective reading in cases where the reader alone is aware of the evidence which led him to draw conclusions. In the case of intersubjective reading on the other hand, the information is shared by a larger group of people (Narrog 2017, p. 23).

From a synchronic perspective, Traugott proposes that “the expressions of subjectivity and (inter)subjectivity are expressions whose prime semantic or pragmatic meaning is to index speaker attitude or viewpoint (subjectivity) and speaker’s attention to addressee’s self‑image (intersubjectivity)” (Traugott, 2010, p. 32).

Narrog (2017, p. 38) argues that subjectivity and intersubjectivity are features of context rather than of specific linguistic units and thus proposes a more general cover term, namely speech act‑orientation, which “encompasses increased orientation toward all the participants in the speech act”. According to this classification, there are three main participants in the speech event: speaker‑orientation refers to the speaker’s stance or perspective towards the situation; hearer‑orientation covers general attention towards the addressee, thereby widening the restricted reference to his self‑image and face needs, as suggested by Traugott; discourse‑orientation describes the way the speaker perceives the interrelation between the various parts of the discourse.

Of the four approaches to (inter)subjectivity, we will show that the findings of the analysis are best accommodated within Narrog’s more general framework. The analysis will show that the use of these adjectives and particularly that of tov and bon is directed at a hearer and uttered for his benefit and must therefore take into consideration his standpoint and perspective. Furthermore, the notion of discourse‑orientation will prove essential in the description of the textual functions of tov and bon as it allows a description of the way the speaker perceives the relation between the different parts of the speech event. It will be argued that the same textual relations are also hearer‑oriented as they are provided by the speaker for the benefit of the addressee in guiding his attention, and thus cannot be accounted for solely within Traugott’s more restricted concept of intersubjectivity, which relates mostly to the addressee’s self‑image.

3. Analysis 

3.1 Hebrew

In this section we will follow the development of expressions of favorable qualities in Hebrew. The discussion focuses on the lexeme tov ‘good’ and briefly compares it to the following lexemes regardless of their gender and number inflections: na’e ‘handsome, pleasant, fair’, yafe ‘beautiful’, hagun ‘decent’, nadiv ‘generous’, mehubad ‘honorable’, ra’uy ‘deserving’. The analysis is based on the conventional classification of the history of Hebrew (Magid 1984), which corresponds to the following corpora:  

• Biblical Hebrew (1300BCE‑200BCE): Online Responsa Project.8

• Sages of the Mishna and the Talmud (200BCE‑ 600CE): Online Responsa Project.9

• Medieval Hebrew (600CE‑1800CE): Ma’agarim, The Historical Dictionary Project.10

• End of 19th century and early 20th century (Revival of Hebrew): Historic Jewish Press.11

• Modern Hebrew: Internet sites, corpora of spoken Hebrew.12

Starting with biblical Hebrew, we find only the following three lexemes, tov ‘good’ in (3), yafe ‘beautiful’ in (4) and nadiv ‘noble, willing’ in (5):

(3)

a.

Va‑yar

Elohim

et

ha‑’or

ki

tov

vayavdel

Elohim

and‑see.fut.3sg

God

acc

def‑light

cpl

good

and‑separate.fut.3sg

God

bein

ha‑’or

u‑bein

haxoshex

between

def‑light

and‑between

def‑darkness

‘God saw that the light was good, and he separated the light from the darkness.’ (Genesis 1, 4)

b.

Hithalaxti

lefane‑xa

be‑’emet

u‑be‑lev

walk. pst.1sg

before‑you

in‑truth

and‑in‑heart

shalem

ve‑ha‑tov

be‑’ein‑e‑xa

asiti.

whole

and‑def‑good

in‑eye‑2pl‑poss

do. pst.1sg

‘I have walked before you faithfully and with wholehearted devotion and have done what is good in your eyes.’ (2 Kings, 20, 3)

c.

Va‑yomer

tov

ani

exrot

brit

it‑xa

and‑say.fut.3sg.m

good

I

make.pst.1sg

agreement

acc‑you

‘Good, said David. I will make an agreement with you.’ (2 Samuel 3, 13)

(4)	Hin‑ax	yaf‑a	ra’aya‑ti	hin‑ax
	Thou‑you	beautiful‑sg.f	darling‑1poss	thou‑you
	yaf‑a	ein‑ay‑ix	yon‑im.
	beautiful‑sg.f	eye‑2pl‑poss	dove‑pl.m
	‘How beautiful you are, my darling, how beautiful. Your eyes are doves.’ (Song of Songs 1, 15)
(5)	Lo	yikare	od	le‑naval	nadiv.
	neg	call.pass.fut.3sg.m	more	to‑villain	noble
	‘No longer will the fool be called noble.’ (Isaiah 32, 5)

As is evident from these examples, the expressions tov ‘good’, yafe ‘beautiful’ and nadiv ‘generous’ denoted favorable internal and external qualities in biblical Hebrew. The adjective tov in (3) is unique as it displays three different interpretations. In (3a) it suggests that the outcome is valuable and beneficial while in (3b) it refers more to virtuousness, morality and religious requirements which the Lord demands. In (3c) tov functions as an interpersonal marker, designating acceptance on David’s part, a function which ceases to exist at later periods and returns only in Modern Hebrew.

Between the years 200BCE and 600CE, a period entitled Sages of the Mishna and Talmud, all seven lexemes display the meaning of a favorable quality, as in (6)‑(7), while tov ‘good’ first acquires the meaning of upscaling quantification in contexts which are associated with quantifiable entities such as wages and sums of money, as in (8):

(6)	Kol	exad	ve‑’exad	omer	atsa‑ti	yafa	me‑’atsa‑to.
	Every	one	and‑one	say.prs.3sg.m	advice‑1sg‑poss	beautiful	from‑advice‑2sg‑poss
	‘Everyone says my advice is nicer than your advice.’
(7)	Yesh	la‑hem	al	ma	she‑yismexu	she‑ne’emar	nadiv	lib‑o.
	There.is	to.def‑they	on	what	cpl‑trust.fut.3pl.m	cpl‑say.pass.prs.3sg.m	generous	heart‑3poss
	‘They have what to trust as he is said to have a generous heart.’
(8)	Liyten	saxar	tov	la‑tsadik‑im	she‑mekaymim	et	ha‑’olam.
	Give.inf	wages	good	to‑righteous‑pl.m	cpl‑sustain.prs.3pl.m	acc	def‑world
	‘To give good wages to the righteous who sustain the world.’

Similar instances of tov denoting upscaling quantification are also apparent in Medieval Hebrew, as in (9), while other adjectives still denote only favorable qualities, such as nadiv ‘generous’ in (10) and hagun ‘decent’ in (11):

(9)	Barux	meshalem	saxar	tov	le‑yere’‑av.
	Bless.pass.prs.3sg.m	pay.prs.3sg.m	wages	good	to‑fearing‑him
	‘Blessed the one who pays good wages to God fearing persons.’ (1024)

(10)	Al	shem	she‑lib‑o	nudvo	karuy	nadiv	lev.
	On	name	cpl‑heart‑3poss	generous	call.pass.prs.3sg.m	generous	heart
	‘As his heart is generous he is called generous heart.’ (1100)

(11)	Dayan	she‑’eyn‑o	hagun	ose	din	she‑lo	ka‑din.
	Judge	cpl‑there.is.neg‑he	decent	make.prs.3sg.m	judgment	cpl‑neg	as.def‑judgment
	‘A judge who is not decent judges unjustly.’ (800)

The end of the 19^th century and the beginning of the 20^th century is viewed as the period of the revival of Hebrew, where it first began to be used as a spoken language. During this period, we find for the first time the meaning of upscaling quantification to be used with all the other adjectives in contexts of wages and sums of money, as in (12)‑(15):

(12)	Lo	hefisu	da’a‑to	shel	ba’al
	neg	appease.pst.3pl.m	opinion‑3poss	of	owner
	ha‑bait	be‑nidon	sxar	dira	na’e.
	def‑house	in‑matter	pay.gen	apartment	nice
	‘They did not appease the landlord on the issue of a nice rent.’ (1887)

(13)	Ha‑mor‑im	Tsrihim	lekabel	saxar	hagun	she‑yuxlu	lixyot
	def‑teacher‑pl.m	need.prs.3pl.m	get.inf	wages	decent	cpl‑can.fut.3pl.m	live.inf
	bli	maxsor.
	without	shortage
	‘The teachers should get decent wages so they can live without shortage.’ (1903)

(14)	Saxar	mexubad	ze	ein‑o	kolel	bonus‑im.
	Salary	respectable	this	there.is.neg ‑it	include.prs.3sg.m	bonus‑pl.m
	‘This respectable salary does not include bonuses.’ (1952)

(15)	Yishlax	la‑nu	kol	mosad	sxum	kesef	ra’ui.
	Send.fut.3sg	to‑us	every	institution	sum	money	worthy
	‘Every institution will send us a worthy sum of money.’ (1928)

An interesting development of tov occurred during the period of Modern Hebrew, when it acquired the meaning of intensification:

(16)	Taviu	sandvich‑im,	maim	ve‑’afilu	kafe,	atem	holxim	levalot	sham
	Bring.fut.2pl.m	sandwich‑pl.m	water	and‑even	coffee	you	go.prs.3pl.m	spend.inf	there
	kama	sha‑ot	tov‑ot	ve‑’ein	hafsak‑ot
	a.few	hour‑pl.f	good‑pl.f	and‑there.is.neg	break‑pl.f
	‘Bring sandwiches, water and even coffee, you are going to spend there a good few hours and there are no breaks.’

(17)	Kedey	lehaxin	et	ze	bizbazti	axshav	kama	dak‑ot	tov‑ot
	To	make.inf	acc	it	spend.pst.2sg	now	a.few	minute‑pl.f	good‑pl.f
	‘In order to make it I wasted now a good few minutes.’

(18)	Naxon	loke’ax	kama	sha’‑ot	tov‑ot	lehaxin	ot‑o,	aval	ze
	Right	take.prs.3sg.m	a.few	hour‑pl.f	good‑pl.f	make.inf	acc‑it	but	it
	be’ikar	lexakot	la‑batsek	she‑yitfax	xuts	mi‑ze	ha‑kol	kalei	kalut.
	mainly	wait.inf	to.def‑dough	cpl‑rise.fut.3sg.m	outside	from‑it	def‑everything	easy‑gen	easiness
	‘True it takes a good few hours to make it, but it is mainly to wait for the dough to rise other than that, it is all very easy.’

(19)	Haktsafa	tov‑a	lokaxat	be’erex	10	dak‑ot	tov‑ot.
	Whipping	good‑sg.f	take.prs.3sg.f	about	10	minute‑pl.f	good‑pl.f
	‘Good whipping takes about a good 10 minutes.’

Examples (16)‑(19) demonstrate two realizations of a partially schematic variant of the construction [Tov]. The first realization of this variant appears in examples (16)‑(18). In these examples the pattern kama sha’ot tovot, lit. ‘a good few hours’ consists of the quantifier kama ‘a few’ and an open slot which may be realized through different units of time such as minutes, hours and days. The expression conveys a subjective evaluation that the activity described takes a long and undefined period of time, undoubtedly much longer than expected by the speaker or the addressee. It should be noted that in Hebrew kama is an interrogative marker in the sense of ‘how much’ or ‘how many’, a quantifier as in Ani rotse kama anavim ‘I want a few grapes’ and an intensifier as in Kama xikiti layom haze, lit. ‘how much I waited for this day’. Interestingly, what we see in the example kama sha’ot tovot is that the entire expression denotes intensification while in Kama xikiti layom haze, lit. ‘how much I waited for this day’, only the quantifier kama functions as an intensifier. A second realization of this variant of the construction [Tov] appears in example (19), where the quantifier kama is replaced with a numeral. The main difference between the two realizations is that the first introduces no limit at all whereas the second presents a minimum boundary which is understood to be exceeded. Both variants seem to be productive, as additional conceivable units of quantitative, numerable nature, such as that of weight, are optional.13

These sentences are associated with pragmatic reading which involves the idea that an activity or a state may go beyond the measurement unit indicated and may therefore be in conflict with one's expectations. This results in a variety of illocutionary values which receive further support from the wider context: the list of things that the addressees are advised to bring with them as well as a clear statement regarding the lack of breaks in (16); complaining about the fact that the baking of the cake turned out to be a waste of time in (17); the speaker’s attempt to justify the long wait in (18); the online site which provides recommendations regarding specific instructions so that the inexperienced baker would know how long the task is expected to last in (19). Notice in (19) that the first tov describes the quality of the whipping whereas the second tov conveys a quantitative meaning.

Modern Hebrew displays a variety of discourse functions of tov (Livnat & Yatziv 2003, Ziv 2006, Maschler 2009). The discourse functions of this adjective constitute an important component in its overall description. Nevertheless, since the purpose of this article is not to provide a comprehensive analysis of discourse markers but rather to examine the development of its various uses, we shall describe this function in brief and return to our focus on the diachronic path of development of these expressions. According to Maschler, tov displays several interpersonal functions, such as agreement to an action very much like English okay and acceptance of some state of things as well as textual functions which mark the beginning and the end of a topic or shifts between episodes. Tov in example (20) expresses a request for acceptance on the part of the speaker and in (21) the marking of the beginning of two episodes in a narrative:

(20)	Ani	rotsa	she‑tistakli	al‑ay,	tov	Metuk‑a?	Rak	tistakli	al‑ay
	I	want.prs.1sg.f	cpl.look.fut.2sg.f	on‑me	good	sweet‑sg.f	only	look.fut.2sg.f	on‑me
	ve‑teyad’i	ot‑i	im	mashehu	mishtane	beseder?
	and‑inform.fut.2sg.f	acc‑me	if	something	change.prs.3sg.m	Okay
	‘I want you to look at me, okay honey? Just look at me and let me know if anything changes, ok?’14

(21)	Hanna:	Tov,	az	ani	asaper	et	ha‑sipur	al	ex	niftsati
		Good	so	I	tell.fut.1sg	acc	def‑story	on	how	hurt.pass.pst.1sg
		‘Okay, I will tell the story about how I was wounded…’
	Galia:	okay
	Hanna:	Tov	az	keshe‑hitgayasti	la‑tsava
		Good	so	cpl‑inlist.pst.1sg	to.def‑army
		‘Okay so when I was drafted to the army.’ (Maschler, 2009 p.188)

3.2 French

The analysis of French deals with the adjective bon whose use is compared in some cases to that of the following adjectives: beau ‘beautiful’, honnête ‘honest’, digne ‘worthy’, décent ‘decent’, honorable ‘honorable’. Examples of Old French, Middle French, Classic French and Modern French have been elicited from the following corpora: BFM 2019; Dictionnaire de l’Académie (all editions); TLFi; Linguee.

Starting from the end of 9^th century, we come across a rather religious interpretation of bon and honnête:15

(22)	Buona	pulcella	fut	Eulalia. (Sainte Eulalie v. 1; 9th)
	Good	virgin	be.pst.3sg	Eulalia
	‘Eulalia was a pure virgin’
(23)	La	sue	juvente	fut	honeste	e	spiritel. (Saint Alexis; 11th)
	The	his	youth	be.pst.3sg	honest	and	spiritual.
	‘His youth was honest and virtuous.’

In (22), buona ‘bonne’ signifies a meaning of meeting all the moral criteria in the right way. The adjective bon as well as some nouns and verbs which meant ‘love’, ‘justice’ and ‘honesty’ were perceived at that time as very strong concepts, which in many cases implied powerful emotions leading to extreme conditions that can bring or end life. Similarly, honnête ‘honest’ in (23) may represent in this period the meaning of ‘just’ and ‘virtuous’, about a person who follows the rules of religion and society. However, in some cases it is more difficult to determine whether the meaning of the adjective was associated with a virtuous aspect or simply with a favorable external or internal trait, as tov and bon can collocate with all kinds of subjects: human, non‑human, concrete and abstract as in (24)‑(27):

(24)	a.	Dist	Blancandrins :	« Mult	bon	plait	en	avreiz	(Roland, v.88;12th c.)
		Say‑pst.3sg	Blancandrin:	Very	good	agreement	of.that	have‑fut.2sg.
		‘Blancandrin said: "you will have a very good agreement."’
	b.	Sur leurs	têtes	ils	lacent	les	bons	heaumes	de	Saragosse.	(Roland, v.996)
		On their	heads	they	lace.up‑prs.3pl	the	good	helmets	of	Saragossa.
		‘On their heads they tighten their shining helmets.’16
	c.	Il	a reçu	tant	de	coups	de	bons	épieux	tranchants !	(Roland, v.584)
		He	receive.pst	so.many	of	blows	of	good	spear	sharp.
		‘He has received so many blows from good sharp‑edged lances.’

(25)	Ki	lui	veïst	Sarrazins	desmembrer,	un	mort	sur	altre	geter,	de
	Who	him	see.irr	Saracen	dismember,	a	dead	on	other	throw,	of
	bon	vassal	li	poüst	remembrer.	(Roland, v. 1972)
	good	vassal	him	can.irr.3sg	remember.
	‘Anyone who would have seen him dismember Saracen throwing one dead over another, would have remembered what a good vassal is.’

(26)	Je	vos	donrai	Bon	consel,	se	vos	me	volés	croire.	(Aucassin v.18; 12th‑13th)
	I	you	give‑fut‑1sg	good	advice,	if	you	me	want‑prs	believe.
	‘I will give you a good advice if indeed you are willing to believe me.’

(27)	Et	il	Est	dignes	d’	entre	en	paradis.	(Saint Alexis, v. 173 ; 11th)
	And	he	be.prs.3sg	honorable	of	go	into	Paradise.
	‘He deserves to enter Paradise.’

The examples in (24) denote a favorable quality which is associated with inanimate objects: In (24a) bon refers to a beneficial and sustainable agreement. In (24b) it indicates a quality of protection and security provided by the helmets to the soldiers and in (24c) it refers to sharp edged lances. Example (25) refers to a rather positive human trait of a vassal implying his loyalty to the master. Example (26) refers to the abstract concept of an advantageous piece of advice. Example (27) describes the dignified actions and intentions for which he merits entrance to paradise.

At about the same period, we begin to see cases where bon appears in contexts of quantifiable entities resulting in a difficulty to distinguish between the notions of quality and upscaling quantification:

(28)

a.

Nus

hom

n’avoit

si

boene

Grace.

(Érec et Enide, v. 2209).

No

man

have‑pst.3sg

so

good

Grace.

‘Nobody possessed such/as much good grace.’

b.

Bon

quinze

jorz

ou

plus

tot

plains.

(L’Estoire de la Guerre Sainte, v. 7205‑7210).

Good

fifteen

days

or

more

all

Full.

‘15 full days.’

	c.	Un	bon	coup	de	poing	(TLFi, 1664).
		A	good	blow	of	fist.
		‘A good punch.’

	d.	Pus	pren	bon	mel,	quit	treis feiz	e	treis feis	escumé,	bone quantité
		Then	take‑imp	good	honey,	cook‑ptcp	3 times	and	3 times	skim‑ptcp,	good quantity
		e	si	il	est	gutus,	pren	le	jus	de	neire
		and	if	he	be‑prs.3sg	gout,	take‑imp	the	juice	of	black
		mentes	.i.	bone quanteté	e	.ii.	tant	del	jus	de	l'
		mint	.i.	good quantity	and	.ii.	as.much	of.the	juice	of	the
		herbe	yve,	si	medlez	ensemble
		herb	yew,	so	mix‑imp.2pl	together
		‘[...] then take a good quantity of good honey, cooked three times and skimmed three times, and if he has gout, take a good quantity of black mint juice and an equal quantity of yew grass juice, and mix them together’.

	e.	Avoir un bon salaire (Dictionnaire de l’Académie 1718)
		‘To have a good salary.’

	f.	Vous avez une bonne traite à faire. (Dictionnaire de l’Académie 1718)
		‘You have a good distance to go.’

In the above sentences (28a‑f), the meaning ranges from a qualitative‑intensifying to a numerable‑quantitative reading. Some other adjectives also developed the meaning of upscaling quantification at about the same time:17

(29)	Il	fait	une	despense	honorable. (Dictionnaire de l’Académie, 1694)
	He	make‑prs2sg	a	purchase	honorable.
	‘He has big expenses.’
(30)	C’	est	un	beau	mangeur. (Dictionnaire de l’Académie, 1762)
	This	be‑prs.3sg	a	beautiful	eater.
	‘He is such a big eater.’

It seems that other adjectives developed it at a much later stage:

(31)	Vous obtenez des super jeux, un bonus de bienvenue décent […]
	‘You get super toys, a decent welcome bonus […]’
(32)	C'est un salaire honorable, en particulier pour quelqu'un qui a connu dans son enfance une vie difficile (Linguee).
	‘This is an honorable salary particularly for one who experienced during his childhood a very difficult life.’

Around the 14^th century, the meaning of large quantity becomes fully established in contexts of quantitative, numerable or measurable units such as wages and distance.

It should be stated that all the adjectives continue to display their original meaning of a favorable quality today. The new meanings are added gradually and seem to replace the use of other adjectives in specific semantic fields, as in the case of bon, which serves to describe a large sum of money or a large salary.

Starting at the beginning of the 18^th century we see a rise in a unique use of bon, similarly to that of Hebrew tov, serving as an intensifier to designate an extended amount of time, weight and distance as in (33)‑(35):

(33)	Il y a	une	bonne	heure	que	je	vous	attends. (Dictionnaire de l’Académie, 1718)
	There‑is.prs	a	good	hour	that	I	you	wait‑prs.1sg.
	‘I have been waiting for you for a good hour.’
(34)	Ce sont chaque année une bonne cinquantaine de kilos de canettes en aluminium qui prennent le chemin [...] (Linguee)
	‘Every year about a good 50 kilos of aluminum cans are thrown away.’
(35)	Sur la zone de Bozcaada, les conditions de vent annoncées sont démentielles et doivent lever une mer très forte avec des creux de 4 bons mètres. (Linguee)
	‘In the area of Bozcaada, the wind conditions reported are crazy and the sea might rise a good 4 meters.’

In examples (33)‑(35), the combination of a conventional standard measurement unit such as weight, time and distance, a quantifier or a numeral and the adjective bon suggests an approximation of the extent over the indicated measure unit. In other words, the entire schematic expression is used as an intensifier suggesting a subjective approximate evaluation.

This approximate evaluation is associated with a pragmatic interpretation similar to the one inferred from the Hebrew examples. The utterance in (33) may serve as a complaint about the need to wait longer than he expected. Example (34) functions to warn the addressee that a larger amount than one would guess is regularly wasted every year. In (35) the speaker warns the addressees that the sea waves are expected to rise to at least 4 meters.

This use of bon seems to display the same syntactic restrictions as its Hebrew counterpart: it is exclusively associated with the adjective bon, the slot of the quantifier can be a specified number or the expression quelque and the slot of the unit is limited to that of quantitative measurement units.

At about the same time as the meaning of intensification of bon evolved, we also see a development of the function of a discourse marker. Bon is considered by some researchers a discourse marker, a textual marker, a pragmatic marker or a discourse particle. In addition, Hansen (1995) defines such expressions as metadiscourse markers and Dostie & Pusch (2007) refer to them as connecters, or even oral markers depending on their use in the conversation and the theory adopted. Some scholars also consider bon as an opening and closing oral marker (Lefeuvre 2011), as a conversation structuring marker (Auchlin 1981) and as un petit mot ‘a small word’ when examining the function in the organization of discourse activity (Brémond 2002).

A variety of discourse functions of bon can be found in dictionaries. According to the Dictionnaire de l’Académie (1694, 1^st ed.) for example, one may reply bon and sometimes bon bon when wishing to express consent and approval. In contrast, one may reply bon also when resorting to irony or mockery if you are told that someone is angry with you. Only in the 3^rd edition of the dictionary do we first see the use of bon as a discourse marker functioning to terminate speech:18

(36)	Bon ! Ah bon ! Allons bon ! C’est bon ! (Dictionnaire de l’Académie, 1740)
	‘Well! Oh well! Ok then! That’s good!
(37)	C’est bon, je m’en souviendrai. (Dictionnaire de l’Académie, 1878)
	‘That’s good, I will remember this.’

The two functions have been used continuously in French in the sense of approval, as in (38), and in the sense of textual marker, as in (39):

(38)	À chaque phrase il disait : Bien, bien. Quand je suis arrivé au corps étendu, il a approuvé en disant : Bon. (TLFi 1942)
	‘At each sentence he said: Well, well. When I arrived at the lying body, he approved by saying: Good.’
(39)	Ça va, c'est bon, intervint la mère, sentant que la dispute tournait à l'aigre. (TLFi 1935)
	‘Alright, it’s ok, the mother intervened, anticipating that the argument was going to end badly.’

Modern use of bon as a marker of shifts between episodes is illustrated in (40) and as a signal of agreement or acceptance in (41):

(40)	Bon, continuons les présentations. (Linguee)
	‘Okay, let’s continue the presentation.’
(41)	Bon, ça me plaît bien tout ça. (Linguee)
	‘Okay, I like it all.’

4. Discussion

4.1 Path of development

The analysis above illuminates several important points regarding the development of favorable adjectives in Hebrew and French. With respect to Hebrew, it appears that tov acquired the broadened meaning of ‘upscaling quantification’ already at the time of the Sages of the Talmud and Mishna about 200BCE‑600CE, while the other Hebrew adjectives only acquired it much later, at the beginning of the 19^th century. Attestations of discourse meaning of tov are evident in biblical Hebrew, a function which ceased to exist for a very long time and showed up again only quite recently in Modern Hebrew. The intensifying meaning also begins to be attested only in Modern Hebrew. In contrast, the other adjectives (e.g. decent, honorable) neither acquired the meaning of intensification nor that of the discourse functions.

The stages of development of the French adjectives seem to be much more condensed. First occurrences of bon as a marker of upscaling quantification are observed as early as the 12^th century. Some adjectives denoting favorable qualities (e.g. honorable, beau) gained this meaning at a slightly later stage. Both the function of intensification and that of discourse marking of bon appeared at about the beginning of the 17^th century.

4.2 Description of the process

Since the beginning of the study of grammaticalization it was suggested by Meillet (1905, 1912) that a certain process may be observed to occur at different stages in the history of languages. Furthermore, as stated by linguists (Lehmann 1985, Hopper & Traugott 2003), a language might skip a stage along its path of development and move towards the next one. In our case, we can see that the evolution of the meanings of the constructions [bon] and [tov] as formulated in Section 1 is parallel in terms of the various stages which they follow despite differences in the time frame and the pace of the changes.

In the first stage, both adjectives denote favorable qualities of behavior, morality, virtuousness and social function, both human and non‑human, as in examples (3)‑(5) and (22)‑(26). In languages of a monotheist religion, it is clear to the hearer or reader that such qualities are considered advantageous and praiseful since the persons in these examples are obedient to the laws of the Lord.

In the second stage, the polysemy of bon and tov is apparent as we see a shift to the meaning of upscaling quantification as demonstrated in examples (8)‑(9) and (28). Unlike the first meaning, which attributes a positive quality to any entity, the new meaning describes quantifiable entities. This expansion of context seems to function as a bridge between the qualitative/evaluative use and the later intensifying use by referring to features such as form and size and shape. Rather than focusing on the qualitative essence of a noun as in the examples of the first type, the sequence 15 jours entiers in (28b) for example, highlights the schematic quantitative characteristics of the day, in this case its length, extending from dawn to sunset. The process of shifting from quality to quantity is observed to have taken place in both languages and for all the adjectives analyzed above.

The third step in the development of tov and bon is the quantitative intensification, as in examples (16)‑(19) and (33)‑(35). The shift from a quantifiable entity to a standard measurement unit together with the shift from a determiner to a numeral/quantifier suggest that the time, weight or distance are now intensified and significantly exceed the frame stated in that unit. Such subjective approximation seems to draw the hearer’s attention to the gap between what is expected and what happens in real life. This interpretation relies on pragmatic factors as it often derives from contexts of recommendations, warnings, complaints or some unexpected results. Furthermore, as the measurement unit is intensified, the hearer reaches the understanding that more effort is involved in completing the task.

We propose that this idiomatic usage may be motivated by speaker‑ and hearer‑orientation. Speaker‑orientation is conveyed in the very subjective evaluation expressed in this use, which overrides the more neutral meaning of the original quantifier. Hearer‑orientation is expressed in the very function of warning or recommending, which are uttered for the benefit of the addressee out of an assessment of his needs.

We suggest that the emergence of the intensifying function is a result of a grammaticalization process which was driven by speaker’s subjectivity. The expression of subjectivity regularly leads to a shift towards more abstract and procedural meaning. In our case the adjectives bon and tov, which regularly denote a favorable quality, undergo decategorialization in terms of Hopper (1991): a shift from a propositional meaning to a pragmatic function which encodes the speaker’s attitude and from concrete to abstract meaning. In fact, the components of the construction are no longer interpreted compositionally and instead, the entire sequence now functions as a quantifier. This process supports previous research on the development of various intensifiers where propositional lexical items become procedural abstract intensifiers (e.g. Shefer & Bat‑Zeev Shyldkrot 2020, Bat‑Zeev Shyldkrot 2001, Athanasiadou 2007, Ghesquière & Davidse 2011). In addition, we observe greater internal dependency between the quantifier, the measurement unit and the adjective. This dependency is realized in the exclusivity of the adjective tov or bon and the syntactic restrictions on the realization of the measurement unit and the quantifier. In sum, the adjectives tov and bon have undergone grammaticalization in developing from a loose sequence to a tightly bound sequence, thereby functioning as a form‑meaning pairing with a new associated meaning.

As for the fourth stage of development of tov and bon, namely discourse markers as described in examples (20)‑(21) and (36)‑(41), we suggest that this evolution is motivated by the relation between hearer and speaker on the one hand and the relation between the discourse and speaker on the other. We adopt Narrog’s general concept of speech‑act orientation as it includes all the functions of tov and bon. Hearer‑orientation incorporates tov and bon’s function as expressing acceptance and approval for the benefit of the addressee. Discourse‑orientation refers to the function of designating a beginning or an ending of an episode. This function also reflects speaker‑orientation in directing and controlling the addressee’s attention and focus.

In order to describe the process that tov and bon underwent in acquiring a discourse function, we consider several features which regularly derive from a grammaticalization process of discourse markers. It appears that tov and bon correlate with the following features: Their scope expands over discourse and they tend to occupy different syntactic positions; they display a number of pragmatic uses motivated by speaker and discourse‑orientation; they are non‑compositional and their meaning is thus procedural rather than conceptual‑propositional (Brinton 2008, Heine 2013 among others).

Regarding the general direction of development of tov and bon, it seems that present‑day polysemy of both is an outcome of parallel influence of speaker‑orientation and hearer‑orientation. As for Hebrew, its unique status as a revived language may be able to account for this parallel. Since Hebrew served for communication at the time of the bible, we would expect a communicative function to exist and be documented at that time. However, as early as 200CE, Hebrew ceased to function as a spoken language for a period of 1600 years, during which it was only used for prayer, bible commentary, administrative books and other formal records, all of which serve as a basis for the relevant corpora used in this study. Therefore, hearer‑orientation or discourse‑orientation functions which are typical of everyday interaction would not have been expected to arise. Only at the time of its revival, at the end of the 19^th century, when the language was brought to life as a fully spoken language, an outburst of developments occurred which led to a simultaneous growth of speaker‑, hearer‑ and discourse‑oriented meanings and functions.

With respect to the direction of development of bon, it appears that speaker and hearer‑orientation have triggered the rise of the intensification function as well as that of the discourse marker at about the same time. Here too, the tendencies seem to be parallel rather than sequential. Whereas subjectivity or speaker‑orientation has been observed in some studies to precede hearer and discourse‑orientation (Traugott 2010), Narrog (2017) suggests that no fixed order can be established and that other directions of development are also possible.

Speaker and hearer orientation are also expressed in the following pair of sentences where the interpretation of a favorable quality of bon depends on pragmatic factors such as the identity of the speaker:19

(42)	J’ai acheté cette voiture à un bon prix.
	‘I bought this car for a good price.’
(43)	J’ai vendu cette voiture à un bon prix.
	‘I sold this car for a good price.’

Clearly, when the speaker is the buyer, a good price would mean a low price. However, when the speaker is the seller, a good price would mean a high price. These examples do not seem to be a case of zeugma, as the two interpretations are not conceptually different (as they are in ‘John lost his coat and his temper’). Rather, both are interpreted as ‘advantageous’ but they represent two different standpoints. As the resolution of this ambiguity appears to be based on pragmatic rather than semantic considerations, there is reason to assume that the two meanings are represented as part of the same concept in the speakers’ lexicon.20

The discussion of the Hebrew and French adjectives raises an important question regarding the unique development of tov and bon compared to the other adjectives as to why only these two adjectives evolved further to function as an intensifier and a discourse marker. One possible answer relates to the frequency of those terms. As we can see on Google Ngram,21 both are much more common compared to all the other adjectives discussed above:

5. Conclusion

The purpose of this paper was to analyze the way in which some adjectives of favorable qualities have come to express present day polysemy. In trying to define the entire process that those adjectives followed, it became clear that due to layering, each stage displays different characteristics and therefore had to be analyzed separately despite their strong interrelation.

The question remains whether the overall development of the constructions [tov] and [bon] whose formulation was described in Section 1, is a case of grammatical or lexical constructionalization. We would like to suggest that an overall grammaticalization path may be identified as follows:

Qualitative marker > quantitative marker > intensifier > discourse marker.

We propose that the shift from left to right can be described as a shift from contentful open category form to a procedural closed category form. This shift is accompanied by increasing degrees of non‑referential, schematic and procedural meanings and is associated with decategorialization, bleaching and layering, all of which are characteristic of grammaticalization. In other words, the adjectives have changed from signaling content to signaling linguistic relations and perspective.

At the same time, we adopt Brinton & Traugott’s definition (2005, p. 89), according to which lexicalization is an “institutionalized adoption into the lexicon”. The lexicon is viewed as an inventory of both lexical and grammatical items. Adoption may be from any layer of language, including regular processes of word formation as well as grammaticalization processes. The rationale behind this view is that items which derive from any of these processes are form‑meaning pairings that are stored in memory and have to be learned (p. 90). It appears then that present‑day polysemy of bon and tov is the output of a grammaticalization process, but that in each stage a new form‑meaning pairing be it a lexical item, or a procedural item has been lexicalized and added to the mental storage.

In a further study that we hope to conduct, we intend to examine the extent to which the meaning of intensification is productive and to look at some additional possible variants of the construction. It may be the case that expressions such as deux heures entières ‘whole two hours’ and deux longs mois ‘two long months’ and their Hebrew counterparts followed the same course as tov and bon although at a different rhythm.