Tracing Early Texts. A Linguistic and Historical Inquiry into Textuality

Abstract This paper tries to identify some of the earliest texts in human history. For this purpose, a historical overview of writing is presented to establish the context in which texts emerged. The subsequent section presents three definitions of text: a minimal definition allowing for various writings to be considered texts if they are cohesive, a second one defining text as unified and coherent composition serving a communicative purpose, and a narrow definition that consists of seven criteria for identifying text: 1) cohesion, 2) coherence, 3) intentionality, 4) acceptability, 5) informativity, 6) situationality, 7) intertextuality. Several critical aspects of these definitions of text, such as the utility of pragmatic factors for identifying texts as such, are discussed shortly. With the three definitions in place, the paper analyzes examples of early text candidates. It concludes that there is no singular answer to the question of the earliest texts in history. Nevertheless, at least one contract from 3100 BC in what is now Iraq is identified as an early piece of writing that can be spoken of as a text.

Introduction

While we can find a trove of literature on the history of writing (Gelb 1952, Gaur 1992, Fischer 2001, Rogers 2005, Schmandt-Besserat 2014), there are surprisingly few to no works explicitly dealing with the history of text.\(^1\) There is no established answer to the question: What were the earliest texts ever? For example, the results from this question being submitted to search engines such as Google or Bing tend to conflate the meaning of writing, text, and literature. Some results show that Sumerian scribes from 3200 BC wrote on clay tablets in cuneiform, most likely ancient shopping receipts (Alex 2019). Another result leads to the Kesh Temple Hymn and the Instructions of Shuruppak from around 2600 BC, which are considered to be an ode and a piece of advice (Andrews 2023).

But were these the firsts texts to be properly spoken of? The goal of this paper is neither to identify early writing nor early literature. Instead, I explore which texts were first created by humankind. However, a lot remains open to what we define as text. It is demonstrated throughout the paper that a plurality of definitions of text also determines the answer to the previous question. It is concluded that we cannot settle on one document as the earliest text in history. That said, an administrative piece from present day Iraq dating to 3100 BC suggests that among the earliest texts were contracts.

This paper contains the following sections: I approach text by recapitulating the history of writing because it provides us with the context from which texts have arisen. In the following section, I present three definitions of text: 1) the minimal definition that allows for many writings to be texts, as long as they are cohesive, 2) the most common definition that sees texts as a unified and coherent composition of sentences, serving a communicative purpose, 3) a narrow definition by de Beaugrande and Dressler (1981), who establish seven criteria for identifying text as such. Their criteria are the following: 1) cohesion, 2) coherence, 3) intentionality, 4) acceptability, 5) informativity, 6) situationality, 7) intertextuality. A short discussion of some shortcoming of these definitions is presented next. I discuss the length of texts as well as the utility of pragmatic features for identifying texts. Finally, I round up this paper with analyzing two pieces of writing. The first is an inscription on a vessel from 3150 BC, and the second is a clay tablet with a narrative about a slave trade that involved multiple parties from 3100 BC. These writings are among the very first in human history.

Recapping the history of writing

Before looking at the historic development of writing, we need a better understanding of what writing is.\(^2\) To achieve this, it may be advantageous to divide the writing-history community into two distinct camps, each of which will provide us with an understanding of writing: There are 1) those, who have a loose definition of writing (Gaur 1992, Schmandt-Besserat 2014), and 2) those, who have a strict one (Fischer 2001, Rogers 2005). The loose camp sees writing as a means of storing and communicating information. Gaur proclaims: “All writing is information storage.” (Gaur 1992: 15) The strict definitions of writing tie the concept further to “the systematic arrangement of significant vocal sounds” (Fischer 2001: 12). So, the strict understanding of writing sees it as the representation of actual spoken language whereas the loose definition would allow for many representational means to be writing.

It is generally agreed by both camps of writing historians that writing has been separately invented in three regions: the Near East, China, and Mesoamerica. However, the earliest traces of writing can be found in Mesopotamia, that is, present-day Iraq, dating back to at least as early as 3200 BC (Schmandt-Bresserat 2014). The invention of writing in this region is also well documented, which is the reason why most scholars focus on the Sumer’s early writing systems, who inhabited Mesopotamia during this period.

Schmandt-Bresserat divides the overall development of writing in Mesopotamia into four phases (2014). The first period of storing information began with clay tokens (8000 - 3500 BC). During the following period, these tokens were translated to two-dimensional symbols (3500 - 3000 BC). This change of making information accessible over time and space is being followed by the first systems of phonetic signs (3000 - 1500 BC), which finally led to the alphabet. The alphabet is a very economical way of representing vocal sounds.

Among the first examples in the strict sense of writing were inscriptions on seals, and vessels which were found in the remains of tombs. These inscriptions did not deal with any merchandise in terms of numbers, but bore names phonetically written. With the ascendance of names, more writing examples emerged in funerary contexts: Status depicting descriptions of individuals were successively added as well as pleas (Schmandt-Besserats 2014). This process eventually led to documents of religious or royal nature, as the Kesh Temple Hymn from around 2600 BC.

Three definitions of “text”

To identify early writings as texts it is necessary to define what constitutes a text. That is, we need to understand the structure of text. Therefore, I will proceed with three definitions of text: the first of which is mostly text internal, that is, it focuses on features that are materially evident in the text. The following definitions increase in complexity and gradually incorporate text external aspects as well. By text external I mean pragmatic features, such as acceptability within communicative situations. For now, this discussion leaves out further dimensions that should ultimately be also considered, such as the media that have been used for text production and dissemination.

The reasoning presented in this paper is guided by a certain philosophical rigor. The definitions offered are intended to serve as conceptual foundations against which specimens of texts will subsequently be compared. Each definition is formulated so that its elements are both necessary and sufficient to identify a piece of writing as a text. Should any definitional condition fail to apply, it is assumed that the writing in question would not qualify as a text under that particular definition. Whether the authors whose definitions are employed here would themselves insist that their criteria must always be applied to determine textuality remains somewhat uncertain, as only Linke and colleagues (1996) have explicitly addressed the intended strictness of their formulations. Accordingly, I invite the reader to regard the following definitions as idealized instruments for the purpose of investigating the earliest texts in human history.

1st definition: Halliday and Hasan define text as follows: “’Text’ is used to refer to any passage, spoken or written, of whatever length, that does form a unified whole” (1976: 1). This definition is being accompanied by an intuitive one. They claim: “We know, as a general rule, whether any specimen of our own language constitutes TEXT or not” (Halliday and Hasan 1976: 1). What matters is that text in its most common form should be distinguished from “a collection of unrelated sentences” (Halliday and Hasan 1976: 1). But that is not to say that any text must consist of at least two sentences. A text is rather “a semantic unit” that can also be found in some proverbs or announcements (Halliday and Hasan 1976: 294). But what constitutes its unity?

For Halliday and Hasan (1976), unity in texts is produced through cohesion. Roughly put, cohesion is what connects a set of descriptions, appeals, and/or propositions. Consider their example: “Wash and core six cooking apples. Put them into a fireproof dish.” (Halliday and Hasan 1976: 2) These sentences form a unit because they are interlaced through the anaphoric pronoun “them”. We can only make sense of the latter sentence by linking the pronoun to the apples from the previous instruction. In this manner, the two sentences constitute a cohesive text.

For Halliday and Hasan (1976), cohesion is not only produced with referential expressions, but also through further linguistic devices, such as repetition, substitution, ellipsis, conjunction, and lexical cohesion. For example, sentences can be linked by having several statements that circulate around the same object, let us assume this would be “car”, which is being referred to either repeatedly in another sentence or by a substituting expression, like “vehicle”. In this case, we would perceive unity in multiple clauses or sentences by identifying the same object at hand. Another device for cohesion according to Halliday and Hasan (1976) is the ellipsis. Consider the following example: “Ida orders a scoop of chocolate ice cream. Ada orders two.” These sentences constitute a unit because we assume that the latter also refers to ice cream, albeit the fact that nowhere it is stated that “two” would imply two scoops of chocolate ice cream. Cohesion through conjunction is created with connecting expressions such as “if… then…”. They explicitly guide the text recipient in putting propositions into relation. Finally, Halliday and Hasan argue that cohesion also results from lexical proximity of words, that is, words such as “people”, “person”, “man”, “woman”, “child”, etc., which fall into the same category of a general class of human nouns (Halliday and Hasan 1976: 274). One sentence may mention one of these, while another sentence may address a lexically related one, allowing the reader to connect the sentences through such lexical proximity.

Linke and colleagues (1996) emphasize that cohesion is necessary for identifying a text, but not sufficient. According to their view, cohesion only covers the superficial structure of texts, that is, the features that are materially evident in clauses and sentences. To differentiate a text from a loose collection of sentences, we would further need an understanding of the underlying deep structure.

It is suggested that we have text if its structure is also coherent (Linke et al. 1996). Coherence is a result of everyday knowledge or supra textual knowledge, which allows the reader to organize the textual elements into a meaningful whole. Consider the following example: “Hans won’t come to the conference. He’s sick.” (Linke et al. 1996: 225, my translation) Although the pronominal expression “he” connects these sentences cohesively, Linke and colleagues believe that its unity is also constituted by our need to relate its parts according to the most probable meaning that we can assign to them (Linke et al. 1996: 225). They argue that we would understand the two sentences in terms of a causal relation: Hans won’t come to the conference because he is sick. This is so because we read such sentences against the backdrop of our world knowledge where a causal relation of two such events would be the most coherent explanation. This will bring us to the second definition of text.

2nd definition: “A text is a complexly structured, thematically and conceptually interrelated linguistic unit with which a speaker performs a linguistic action with recognizable communicative meaning” (Linke et al. 1996: 245, my translation).\(^3\) When Linke and colleagues speak of “structured, thematically and conceptually interrelated linguistic unit”, their notion of unity should be explained with cohesion as well as coherence. For example, unity may be created with certain structures, such as expected behavioral patterns that are manifest in the text composition, or in terms of conceptual proximity, for example by using lexically cohesive words. A theme running through the text would be one explicit dimension which creates coherence; it is another means for the recipient to see the text as a meaningful whole by identifying an idea that pervades the text.\(^4\)

Linke and colleagues (1996) add that texts have a pragmatic side. A text must also be seen as “a linguistic action”. According to their definition (Linke et al. 1996: 246), at least three actions would be identifiable that can be realized using texts, fulfilling the following communicative functions: 1) texts describe objects, situations, or events; 2) texts express the emotions and intentions of their authors; 3) texts are appeals to their recipients with the aim to produce a particular reaction. For instance, a construction manual can contain texts, which describe actions that must be performed to build an object. A love letter expresses its authors devotion to someone. The “park rules” of Potsdam’s Park Sans Souci are meant to prevent visitors from behavior that would damage the property or annoy fellow visitors. In this manner, the text appeals with the aim to elicit a certain behavior.

3rd definition: De Beaugrande and Dressler (1981) propose the most complex definition of text. They argue that “text is a communicative occurrence that meets seven criteria” (de Beaugrande and Dressler 1981: 3, my translation). These criteria are the following: 1) cohesion, 2) coherence, 3) intentionality, 4) acceptability, 5) informativity, 6) situationality, 7) intertextuality. The authors are convinced that they can be spelled out for each text.

I will only briefly summarize 3-7 because de Beaugrande and Dressler’s understanding of cohesion and coherence resembles the previous explanations of these concepts. (3) By mentioning intentionality in their definition, de Beaugrande and Dressler also emphasize the speech-act-theoretical side of texts. The purpose of each text is to achieve a specific objective, such as commemorating someone (de Beaugrande and Dressler 1981: 9). (4) Their notion of acceptability is supposed to refer to the attitudes of text recipients. The idea is that texts can only be perceived as such when they are aligned with the expectations and attitudes of their addressees (de Beaugrande and Dressler 1981: 9). (5) Closely related to acceptability is the concept of informativity. For de Beaugrande and Dressler, texts would always be informative. To what degree a text is informative can be measured by pondering the following questions (de Beaugrande and Dressler 1981: 10). Is the text expectable? Are the text’s elements unknown, partially known, or known? Are any of the text’s elements improbable or probable? (6) Situationality is about what makes a text relevant to a particular conversational goal (de Beaugrande and Dressler 1981: 12). In that, it can be seen as another guiding principle, helping to realize the text’s intention(s). Situationality is especially relevant when we consider the context of text. Finally, (7) de Beaugrande and Dressler introduce intertextuality as a relevant criterium for texts. Its notion is tied to informativity and acceptability. Intertextuality describes the degree to which a text requires prior knowledge of other texts in order to be acceptable to its intended audience and to be appropriate for use in a given context (de Beaugrande and Dressler 1981: 13).

By mentioning intertextuality, de Beaugrande and Dressler (1981) emphasize the possibility of separating texts into distinct categories. They speak of text sorts as “a class of texts that are expected to have certain properties for certain purposes” (de Beaugrande and Dressler 1981, 188, my translation). However, they abandon the pursuit of precise classification of texts, arguing that a typological approach must remain vague due to the presence of too many indeterminate examples of texts (de Beaugrande and Dressler 1981: 193). To get an idea of the difficulty of classifying texts, Dimter (1981) has created a non-exhaustive list of possible text sorts, with at least 500 different text type designations that can be found in German. That said, linguists, such as Sandig (1972), have nevertheless attempted to come up with taxonomies.\(^5\) While the limit of these now appears to be broadly accepted, non-exhaustive text classification has gained new popularity in Computational Linguistics since the emergence of classification tasks as a dominant theme in machine learning (Stede et al. 2006).

Some caveats to the definitions

Having discussed three definitions of text, which are increasingly complex, we need to consider how these are helpful with identifying historic writings as texts. The definitions provided by 1) Halliday and Hasan (1976), 2) Linke and colleagues (1996), as well as 3) de Beaugrande and Dressler (1981), appear to have all their advantageous and disadvantageous. I have already mentioned Linke’s critique of Halliday and Hasan’s view. The remainder of this section is limited in scope and may only address a few further critical points. I would like to discuss size of texts, the limits of speech-act-theoretical aspects for identifying early texts, and the utility of some of the criteria by de Beaugrande and Dressler (1981), such as situationality.

Whether a few words, a clause, a sentence or several ones, an inscription, or a collection of lines may be considered as text is also tied to the question how large a text is. Is a word text? Or do we need two words, noun and verb, to deal with text? To what degree does size matter? Halliday and Hasan’s definition of text deviates from the notion that text is something bigger than a sentence. For they believe that proverbs like a picture is worth a thousand words could be interpreted as text. However, many scholars from antiquity until today have defined text as something that follows the sentence in a hierarchical order. For example, Scherner (1996) argues that Plato had already identified text as the ultimate unit of language, which is larger than a sentence, with the caveat that Plato used the Greek word lógos (λόγος) to speak of texts (Scherner 1996: 106).\(^6\) Heringer, a contemporary scholar, likewise views text as the entity that succeeds the sentence in size (Heringer 2015: 11).\(^7\) It is worthy keeping in mind what size we settle on may also decide whether some writings can be termed text.

Linke and colleagues (1996) as well as de Beaugrande and Dressler (1981) highlight extra textual, that is, pragmatic features of text. I wonder to what degree some of these features are relevant for identifying early texts in history due to our partial knowledge about concrete customs, worldviews, and cultures of past times. What if we cannot determine the intention of an ancient piece of writing? Early literature, such as the Kesh Temple Hymn, might have served religious, but also magical beliefs of the time that we do not know anymore or for which we have no proof. For these matters, we might also misinterpret some texts. Some phrases could seem non-sensical to us, although they had transported a coherent message for the people of their times. Perhaps they carried more than what we can read today.

In similar fashion, de Beaugrande and Dressler’s notion of situationality can be questioned. While it can help to understand the structure of those texts confronting contradictory or unexpected perspectives, it is doubtful whether all texts require situationality for them to be intelligible. This will be more concrete when we think of the examples of traffic control signs given by de Beaugrande and Dressler such as “Slow down. Kids playing.” (1981: 12) The sentences make especially sense as soon as we imagine ourselves seeing the sign next to a busy street while driving a car. We can say that the situation would be conducive for our understanding. However, if we were reading it within a theatre, we would still be able to discern its message. We only seem to be at odds if any hypothetical context for a given piece of communication is foreign to us. However, if that is the case, we may already encounter issues with the actual semantic content of such statements, which would be more fundamental than considering the situation in which a message is intended to communicate. Is situationality then a necessary requirement for identifying texts?

When pondering the need of situationality for text identification, we may also ask whether texts would always be inter-textual: It may sound simple, but identifying the earliest text implies that there is no canon of other texts to which it can be related. Thus, according to de Beaugrande and Dressler, this would suggest that the first recipients of text were unaware of the fact that they were reading text.

Identifying early text candidates

To discuss which pieces of writing are text, I will present two of the earliest discovered writing specimens in human history. These are taken from Woods et al. (2015), who present a systematic and chronological overview of early writing. The first specimen is an inscription and the second an administrative piece.

1.) The funeral inscription is from Egypt, Abydos, Umm el-Qa’ab, ca. 3150 BC, Dynasty I reign of Djer, Egyptian 0 period. It is written on a cylindrical vessel that was found at the tomb of Djer. It reads:

ḥtp-Nἰt

which can be translated as “May Neith be satisfied.” (Woods et al. 2015: 128) The inscription is a name, bearing a reference to the goddess Neith; it was popular for royal females of that time (Woods et al., 2015: 128).

2.) The administrative piece is from Iraq, potentially Larsa, ca. 3100 BC, Uruk III period and was found as can be seen in Woods et al. (2015: 80). The piece describes how twelve slaves are being transferred to new owners. The potential text in modern English reconstructed could sound like what follows:

Obverse: Slaves are sold to the cultivators. Their names are Nim, and others. Slaves, who are called En, Bu Du, Gul, are sold to Maran. Via Parmud and via the pig herder of Adab.

Reverse: The subtotal of six slaves via the pig herder of Adab. Subtotal of six slaves via Parmud. Twelve slaves are for Maran and Ente.

It is not clear though what the order of the sides is, and if one should be read before the other.

When discussing these examples, the idea of Halliday and Hasan (1976: 1) to test our intuition about whether a specimen constitutes a text is appealing to me. Intuitively, the shortest piece of writing would probably be the most controversial example since it also defies the common view that text is something more than a single sentence or word. Therefore, we shall begin by examining whether or not ḥtp-Nἰt sufficiently fulfills any of the previously discussed definitions.

The funeral inscription appears to have two semantic components: On the one hand, it is a proper name, on the other, the name has meaning with proverbial traits: “May Neith be satisfied.” If read as proper name, the inscription would not fulfill the definitions of text because it lacks elements that are brought to unity through any of the possible features that text appear to have. But can ḥtp-Nἰt be really read as a proverb? The Merriam-Webster dictionary defines a proverb as “a brief popular epigram or maxim“ (Merriam-Webster 2023). While that leaves room for interpretation, the expression might for example implicitly carry a maxim that the goddess Neith should always be satisfied, my reading of ḥtp-Nἰt would be closer to something like the Arabic salam alaykum (ٱلسَّلَامُ عَلَيْكُمْ), which accompanies greetings. In similar fashion, ḥtp-Nἰt could be an accustomed expression, whose function was less tied to its meaning to satisfy a goddess, but rather a blessing that was used for specific purposes. In the Egyptian context, it might have taken on a new function, by particularly naming aristocratic females with this expression. It might be no coincidence that salam alaykum also reappears in names, such as Jerusalem, which supposedly carries salam. Therefore, if ḥtp-Nἰt was not a maxim, but close to a blessing, such as salam alaykum, we might not want to speak of it as a proverb.

Halliday and Hasan (1976) go as far as saying that even announcements fall into the category of text. For them, this would also include phrases such as “No smoking” or “For sale” (Halliday and Hasan 1976: 294). However, it is challenging to identify the cohesive characteristics of these types of phrases. In similar vein, if ḥtp-Nἰt is interpreted as being similar to an announcement, we would have to ask what its cohesive elements are?

Cohesion of announcements could be derived from context. The phrase “For sale” does make sense to us when read in front of a house. We would see it as a unit that would not require any additional elements to be understood. Likewise, ḥtp-Nἰt may constitute a unit in particular situations where the expression might have been used, like salam alaykum is used in greetings. The problem with deriving unity in this manner would be for the first definition of text that it goes beyond Halliday and Hasan (1976) since they do not consider pragmatics as cohesive factors.

With Linke and colleagues (1996), we may ask: What makes the name – as saying – coherent? Again, it appears that the best shot would be by considering pragmatics. Let us assume that the phrase was not only used in names, but also in situations to express devotion to the goddess, as I speculate. Using it could have been a way to show the authors’ or parent’s commitment to the community. People might have pronounced ḥtp-Nἰt as a statement like “Jesus is your hope!”, which can be read on billboards in the U.S. If that is the case, ḥtp-Nἰt may have been an act of communication within an expected context that was coherent against the backdrop of the community’s values and customs.

However, with our lack of resources regarding the contexts of the people from Abydos, it is difficult to settle these questions. Therefore, I believe that ḥtp-Nἰt is only text when we stretch our interpretation of the inscription to fit the criteria set out by Halliday and Hasan (1976) and Linke and colleagues (1996). We would need to agree that announcements or short statements are text. Then we must decide if names that also have an additional meaning can be interpreted as proverbs or announcements. If none of this applies, we must consider whether a simple name on a tomb, without any additional proverbial meaning, qualifies as a text. And the answer to this consideration would be no, because not every written word constitutes a text – otherwise, the definitions of text would become redundant.

Given that de Beaugrande and Dressler (1981) have the most complex definition, it is reasonable to say that ḥtp-Nἰt would need to clearly fulfill the first two definitions to justify an extensive analysis based on their definition of text.

The second piece of writing deals with the trade of twelve slaves to new owners, some of whom are called by their names. To start our analysis, we should consider whether the passages on the clay tablet create a unified whole through cohesive elements. The most apparent cohesive feature is repetition: The sentences summarize the transaction’s details in multiple ways, such as by introducing additional information about the parties involved in the sale of the slaves. We can also observe substitution. The slaves are being referred to as slaves as well as by their names. In a similar manner, the new owners are mentioned as cultivators and by their names, Maran, and Ente. We may also deal with lexical cohesion. Cultivator and pig herder could be grouped under the hypernym of agricultural occupations. In brief, we have evidence for several cohesive structures.

The sentences are coherent in multiple ways, too. For example, they have a theme, the sale, which runs through the passages, allowing us to consider that the trade must have been mediated by pig herders when we read “via the pig herder of Adab”. Also, that the addition of six slaves adds up to twelve is coherent. We do not know, in what order the tablets were supposed to be read, but we can imagine that the obverse with more details is being summarized by the reverse, which has less information. Someone using the tablets might have looked at the summary to remember the transaction, and if they had forgotten the details, they could have looked them up on the obverse.

Given that we can identify several cohesive as well as coherent elements, the second piece of writing qualifies as text, according to Halliday and Hasan (1976) and Linke and colleagues (1996). Does it also meet de Beaugrande and Dressler’s (1981) standard?

If we take for granted that the slave trade piece is cohesive and coherent, we shall examine the following points in more depth: intentionality, acceptability, informativity, situationality, intertextuality. The intentionality of the slave trade writing could be at least twofold. Its purpose could have been to describe and record the transaction that took place so that the rulers and fellows of the community knew what had happened. The piece could also be of contractual nature. A contract is a declaration, establishing and ensuring (the property) rights of those parties mentioned in the contract. Even though slavery is not acceptable today, it is fair to assume that it was acceptable at the time of the transaction. Since slavery was not uncommon in Antiquity, the content of the tablets was also expectable and the mediation of such a trade by livestock owners probable. Those, who were familiar with the land and community where the trade took place, probably knew the owners and mediators by name, too. In short, we can assume that the sentences on the tablets were also informative to fellow members of their community although they are less so for us.

It is more difficult to determine if the description of the trade fulfills de Beaugrande and Dressler’s notion of situationality. Because we know little of the context of the writing, we are forced to speculate what situations used to accompany it in ancient Iraq. But what we know is that it was probably found around Larsa. At the time of its creation, it might have been stored in an administrative context that allowed the community of Larsa to settle property claims. In this regard, it might have been accompanied by similar pieces that we do not know of, and which also documented similar transactions. In this manner, our specimen could have been associated with these texts, potentially allowing the reader to understand its nature as a contractual settlement, which assured the rights of the mentioned parties.

Since we have evidence that the second specimen analyzed in this paper is cohesive and coherent, and because we can also spell out intentionality, acceptability, informativity, situationality, and intertextuality features, I conclude that the document from Larsa meets de Beaugrande and Dressler’s seven criteria (1981). So that it can be considered an early text in human history given that is from 3100 BC. If it were the case, that the text not only documented a slave transaction, but was also regarded as a binding arrangement, then we can deduct from this instance that among the earliest texts were texts of contractual nature.

On a final note, the possibility that the slave transaction could be a contract should not be interpreted as implying that early texts were the result of accounting systems. The idea that accounting is at the heart of everything is often linked to the evolution of writing and would be true for any broad understanding of writing, such as knotting techniques or clay tokens, which were used to count livestock or goods. But such a mono-interpretation of the origins of writing already crumbles if writing in its narrow sense also emerged through phonetically written names in funeral contexts. I believe that early texts should be neither reduced to accounting. As has been shown in this paper, some of them were documents of early legal structures that allowed the community order.

In sum, this paper explored the origins of the earliest texts in human history. After presenting a historical overview of the development of writing, it proposed three definitions of text, ranging from a broad, minimal definition based on cohesion to a narrower, seven-criterion framework encompassing cohesion, coherence, intentionality, acceptability, informativity, situationality, and intertextuality. Applying these definitions to early examples of writing, the analysis found that while no definitive first text can be identified, the contract from 3100 BC in ancient Iraq meets the criteria of all definitions sufficiently to be considered an early instance of full textuality.

Footnotes

\(^1\) Some of the afore mentioned authors touch upon early texts (e.g. Schmandt-Besserat 2014), but no one exclusively writes about the history of text. Greetham (1999) has a promising chapter called “The History of the Text”, but it deals with the phenomenon of texts being recycled in different literary discourses.

\(^2\) Writing should not be confused with script. Script would be in cuneiform, italic, bold etc. When I refer to writing, I mean an overall system of signs or symbols.

\(^3\) Bußmann’s definition of text is very similar (Bußmann 2002: 683). Also, Brinker’s (2010: 19-20) summary of text characteristics is close to Linke and colleagues (1996).

\(^4\) A theme comprises the “content-semantic guideline and quintessence of the text”, which would be still perceptible when a text is radically cut (Linke et al. 1996: 237). For another discussion of themes in texts, please see Brinker (2010: 11).

\(^5\) Creating a typology of texts starts with the difficulty of finding the right angle from which texts can be classified. Should they be ordered according to a) communicative criteria, b) domains where texts are being used (such as religion, journalism, art, politics, law, science etc.), or c) according to their types of speech acts (informative texts, declarations, appeals, obligations, contacts) (Heringer 2015: 128)? Another attempt of classifying texts could also divide texts according to text internal features vs. text external ones. Internal features could be layout, length, themes, lexicon, grammatical constructions and so on. The text external features could order texts according to situations where they are being used most often. Heringer (2015: 129) suggests that any serious typology of texts should fulfill the following benchmarks: A typology should be complete, selective, and applicable.

\(^6\) Despite the Latin origin of the word “text”, it was only rarely used. Quintilian speaks of text to refer to language being “weaved” together (Scherner 1996: 109).

\(^7\) Gülich and Raible (1977), among others, explicitly agree with Halliday and Hasan that a single sentence can be text (Gülich and Raible 1977: 51).

Bibliography

Alex, Bridget. 2019. What the Earliest Texts Say About the Invention of Writing. Discover. https://www.discovermagazine.com/planet-earth/what-the-earliest-texts-say-about-the-invention-of-writing (accessed 26 May 2023).

Andrews, Evan. 2023. What is the oldest known piece of literature? History. https://www.history.com/news/what-is-the-oldest-known-piece-of-literature (accessed 26 May 2023).

de Beaugrande, Robert-Alain, and Wolfgang U. Dressler. 1981. Einführung in die Textlinguistik. Tübingen: Max Niemeyer Verlag.

Brinker, Klaus. 2010. Linguistische Textanalyse. Eine Einführung in Grundbegriffe und Methoden. Berlin: Erich Schmidt Verlag.

Bußmann, Hadumod (ed.). 2002. Lexikon der Sprachwissenschaft. Stuttgart: Kröner.

Dimter, Matthias. 1981. Textklassenkonzepte heutiger Alltagssprache: Kommunikationssituation, Textfunktion und Textinhalt als Kategorien alltagssprachlicher Textklassifikation. Tübingen: Max Niemeyer Verlag.

Fischer, Steven R. 2001. A History of Writing. London: Reaktion Books.

Gaur, Albertine. 1992. A History of Writing. New York, London, Paris: Cross River Press.

Gelb, Ignace J. 1952. A Study of Writing. Chicago, London: University of Chicago Press.

Greetham, David C. 1999. Theories of the Text. Oxford: Oxford University Press.

Gülich, Elisabet, and Wolfgang Raible. 1977. Linguistische Textmodelle: Grundlagen und Möglichkeiten. München: Wilhelm Fink Verlag.

Halliday, Michael A. K., and Ruqaiya Hasan. 1976. Cohesion in English. London: Longman.

Heringer, Hans J. 2015. Linguistische Texttheorie: Eine Einführung. Tübingen: A. Francke Verlag.

Linke, Angelika, Markus Nussbaumer, and Paul R. Portmann. 1996. Studienbuch Linguistik: ergänzt um ein Kapitel »Phonetik und Phonologie«. Tübingen: Max Niemeyer Verlag.

Merriam-Webster. 2023. “Proverb.” Dictionary, Merriam-Webster. https://www.merriam- webster.com/dictionary/proverb (accessed 26 May 2023).

Rogers, Henry. 2005. Writing Systems: A Linguistic Approach. Malden, Oxford, Carlton: Blackwell Publishing.

Sandig, Barbara. 1972. Zur Differenzierung gebrauchssprachlicher Textsorten im Deutschen. In Elisabeth Gülich and Wolfgang Raible (eds.), Textsorten. Differenzierungskriterien aus linguistischer Sicht, 113–124. Frankfurt a. M.: Athenäum-Verl.

Scherner, Maximilian. 1996. „TEXT“: Untersuchungen zur Begriffsgeschichte. In Archiv für Begriffsgeschichte, 1996, Vol. 39 (1996), 103-160.

Schmandt-Bresserat, Denise. 2014. Writing, Evolution of. In James D. Wright (ed.), International Encyclopedia of the Social Behavioral Sciences (Second Edition), 761-766. Oxford: Elsevier.

Stede, Manfred, Heike Bieler, Stefanie Dipper, and Arthit Suriyawongkul. 2006. SUMMaR: Combining Linguistics and Statistics for Text Summarization. Proceedings of the 17th European Conference on Artificial Intelligence, (ECAI-06), Riva del Garda, Italy: 827-828.

Woods, Christopher, Emily Teeter, and Geoff Emberling (eds.). 2015. Visible Language: Inventions of Writing in the Ancient Middle East and Beyond. Chicago: Oriental Institute of the University of Chicago.