Usor:Robert.Baruch/Latin Word Order

Introduction

These are my notes on the book Latin Word Order. The book is about how semantics (meaning) and pragmatics (the context) affect the ordering of Latin words. While Latin is supposed to be a language with free word ordering, the order in fact conveys information.

I'm no linguist, and have only read a few books about syntax, so none of the information here may be correct.

Methodology

The authors consider various classes such as verb arguments or noun modifiers, taking a representative sampling of examples from the corpora of classical prose authors.

Pragmatic values

The authors adopt a set of pragmatic values which can be assigned to any word or phrase in a sentence. Generally, they are categorized as focus, topic, and tail.

Any given sentence conveys information the reader already knows, and information the reader does not. Every sentence can be considered as the answer to a question that the reader asks. So, for example in "The apple is red," most likely the apple is already known but we are telling the reader that it is red. The corresponding question is, "What color is the apple?"

In, however, "What is red is the apple," most likely the reader already knows that something is red, and probably that there is an apple previously mentioned, but now the reader has the new information that the apple is red. The corresponding question is, "What is red?"

The authors use the metaphor of filing index cards. In "The apple is red," we have an index card for the apple, and with this new sentence we write on that index card that it is red. In "What is red is the apple," we have two index cards, one for the apple, and one for something red, and with this new sentence we can throw away the "something red" card and write on the apple card that it is red. The same result, but through different processes.

Focus is associated with new information, while topic and tail are associated with old information.

• Focus
  • Weak focus
    • Informational focus: new information (a property) about something already known
    • Presentational focus: new information about attendant circumstances
  • Strong focus
    • Simple (exhaustive) focus: new information that allows no additional information of the same type
    • Counterassertive focus: new information countering a previous assertion
    • Contrastive focus: new information contrasting with a previous assertion
• Topic: the part of the sentence that is already known
  • Weak topic: old information that is not contrastive
  • Strong topic: old information that is contrastive
• Tail: old information which isn't a focus.

Some examples follow.

Informational focus

That monkey was hilarious. The word hilarious (or the phrase was hilarious) has informational focus because it is new information about that monkey which is already known. You could ask the question What was the monkey? or What property did the monkey have? and answer with Hilarious.

That monkey grabbed the banana. Here, the question is What did the monkey do? And the answer is Grabbed the banana.

Presentational focus

When that monkey grabbed the banana, I almost died from laughter. Again, that monkey is already known, and maybe we also know that that monkey grabbed the banana (from previous sentences), but that I almost died from laughter is new information about attendant circumstances. You can't ask What property did the monkey have when it grabbed the banana? and answer with I almost died from laughter, which makes this presentational and not informational. You could, however, ask What happened when the monkey grabbed the banana?

The monkey had a trainer following him. The monkey that was previously mentioned doesn't have the property of having a trainer following him. This is again an attendant circumstance. The question could be, What was going on with the monkey? He had a trainer following him.

Simple (exhaustive) focus

That monkey just grabbed the banana and tore off. This is like the above That monkey grabbed the banana, but does not let any other action take place. The only things that happened were that the monkey grabbed the banana and tore off.

It was me who laughed at the monkey. The implication being that nobody else did. In answer to the question Who laughed at the monkey? you could reply I did, although you don't say whether anybody else did, and that would be weak informational focus. But if you reply It was me, you are saying you and you alone laughed, and that makes it strong exhaustive focus.

Counterassertive focus

That monkey WAS funny. Imagining the sentence with emphasis on was, we are contradicting a previous statement that the monkey wasn't funny. Thus, was has counterassertive focus.

No, the monkey was brown with white spots. Here brown with white spots has counterassertive focus, assuming that previously we asserted that the monkey was some other color. If, on the other hand, we previously asserted that the monkey was brown with black spots, then white spots would have the counterassertive focus. Typically in English stress is placed on the counterassertive focus.

Contrastive focus

That monkey didn't take the orange, but the banana. We establish the set of alternatives and choose one, the set here being the orange and the banana. The banana has contrastive focus. Presumably we know the monkey took something already, which is the old information in the sentence.

You laughed at the monkey, not me. You has contrastive focus, the set being you and me.

Weak topic

The monkey ran away. Since we already knew (old information) there was a monkey, all this sentence is telling us (new information) is that it ran away. The monkey is the topic.

What he wanted was a banana. We know he wanted something, so that is the topic.

Strong topic

Monkeys are from Mars, vipers are from Venus. Monkeys and vipers are strong topics because they are being contrasted. Presumably we already know that monkeys and vipers exist.

Tail

The monkey's paw swatted the banana across the stage. While swatted the banana across the stage might be the focus, and The monkey's paw is the topic, the banana and the stage are already known to exist, but they are not the topic. Hence, they are tails.

Structural trees

Structural trees are used in all sorts of linguistic theories to map out which word joins up with what phrase. Also, rules about the order of words become rules about the manipulation of tree elements.

One of the more widely used structures has nodes having either no branches (corresponding to a single word), or two branches (corresponding to a phrase of a sort). Of the two branches, the first is the head, and the second is its complement. Thus, a sentence like The monkey ran up the big tree, we could end up with a tree like this:

       S
      / \
     /   \
    DP    \
   / \     \
  D   N     \
 the monkey  \
             VP
            / \
           /   \
          V    PP
         ran  /  \
             /    \
            P     DP
           up    /  \
                /    \
               D     AP
              the   /  \
                   /    \
                  A      N
                 big     tree

The sentence has a head, the monkey, and its complement is ran up the big tree. In turn, the determiner phrase the monkey has a head, the, and its complement is monkey.

Theories

For each class, the data is fitted to four structural theories, some more successfully than others. The four theories are:

• Minimal structure theory
• Maximal structure theory
• Functional theory
• Prosodic theory

Minimal structure

This theory posits that every phrase consists of a head and its complement, and pragmatic function is not indicated at all. For example, the structure of the Latin phrase matrem Phalaridis would be drawn (badly and in ASCII) as follows:

      NP
     /  \
    /    \
   N      NP
matrem Phalaridis

Maximal structure

Where the minimal structure theory allows a phrase to be broken into two pieces and thus have two positions for each phrase, the maximal structure theory allows as many positions as there are pragmatic values. Thus, the above phrase, assuming that there is focus on matrem, would become:

          FocNP
           / \
          /   \
         NP   NP
     matrem  /  \
            /    \
           N     NP
          ---   Phalaridis

Thus, matrem has been assigned as the head of a focused noun phrase, leaving the head of its complement null.

If, on the other hand, we have Phalaridis matrem and we decide that Phalaridis is the focus, we get this:

          FocNP
           / \
          /   \
         NP   NP
 Phalaridis  /  \
            /    \
           N     NP
         matrem  ---

Functional

In this theory there are only three pragmatic values: topic, focus, and tail. Every phrase must fit into one of these two trees:

          TopNP
         /     \
        /       \
       N        FocNP
       i       /     \
              /       \
             NP       NP
             j       /  \
                    /    \
                   N     NP
                 ---i   ---j


          FocNP
         /     \
        /       \
       NP       NP
       i       /  \
              /    \
             N     NP
           ---i     j

i and j are words, and ---i and ---j show where the words were in the original structure before they got kicked upstairs (the linguistic jargon is raised) due to having pragmatic value. Note that in both cases, i is the head and j is its complement, but in the tree they appear in different order based on what pragmatic values are occurring. In the first tree, i is the topic and j is the focus. In the second tree, i is the focus and j is tail information.

Prosodic

In this theory, word order is dictated by the effect of stress induced by pragmatic value on the default stress patterns in the sentence. The authors don't go into this theory in much detail, and I will not go into at all.

Neutral word order

The authors first examine a wide range of sentences consisting of focus, sentences answering broad questions like What happened? and What did the subject do? From these sentences, which have essentially no messy pragmatics altering their order, they tease out a default phrase order for Latin:

Subj DO IO/Obl Adj Goal/Source Nonref-DO V.

Subj is the subject of the sentence (nominative) and DO is the direct object (accusative). IO is the indirect object (dative), and Obl is the oblique (which is any case except nominative and vocative).

Adj is the adjunct, an adverbial representing circumstances. This includes instrument, cause, time, place, manner, means, and as you can guess deals with the ablative and the ablative absolute.

Nonref-DO is a nonreferential direct object, which is a kind of direct object so tied to the verb that it cannot be separated. Several examples given: impetum facere (to make an attack), where impetum is a nonreferential direct object, gratias agere (to give thanks), where gratias is a nonreferential direct object, opem ferre (to bring assistance), iter facere (to journey), bellum gerere (to wage war). The reason these are nonreferential is that the direct object cannot be separated from the verb when asking a question. We cannot ask What did Caesar make? (an attack) or What did Caesar give? (thanks).

Finally, the verb comes at the end.

Complement syntax

Complement syntax is an ordering where the verb comes first, then follows words which narrow down the meaning. The example given is from the 6th century Vulgar Latin Itinerarium Antonini Placentini:

...omnes fundunt illos colathos in fluvium...
(...they all immerse the baskets into the river...)

Upon first hearning fundunt, we know that they immerse something. Then we hear illos colathos, and we understand that they immerse the baskets. Then we hear in fluvium, and we understand that they immerse the baskets into the river, each phrase narrowing the meaning.

The order of the verb's arguments is in neutral order, it's just that the verb comes first. The structure posited is as follows:

         VP
        /  \
       /    \
      Subj   \
     omnes    \
              V'
             / \
            /   \
           V     \
        fundunt   \
                  VP
                 /  \
                /    \
               NP     \
     illos colathos    \
                       V'
                      / \
                     /   \
                    V    PP
                   ---    in fluvium

Each restriction is the head of its phrase, except for the final one which is a complement of a null. But this doesn't work for Classical Latin because if the verb comes last, the head would have to come second.

Specifier syntax

In specifier syntax, the authors posit that initially the words occur in complement order, but the arguments raise. Also, the rightmost position is focus, and other higher positions are topic. Consider the sentence

scribas anulis in contione donarunt
(they give the scribes rings in the meeting)

We start with a complement structure:

            VP
           /  \
          /    \
         VP     Adj
        / \     in contione
       /   \
      V'    \
     / \     Instr
    /   \    anulis
   V     \
donarunt  \
          DO
        scribas

First we raise the phrase closest to the verb, which is the direct object, and label its node as the focus:

       FocVP
       /  \
      /    \
     DO     VP
 scribas_i /  \
          /    \
         VP     Adj
        / \     in contione
       /   \
      V'    \
     / \     Instr
    /   \    anulis
   V     \
donarunt  \
          DO
         ---i

Next we raise the next phrase, which bumps the direct object up into a higher node, which we label topic:

      TopVP
      /  \
     /    \
  DO      FocVP
scribas_i /  \
         /    \
      Instr   VP
   anulis_j  /  \
            /    \
           VP     Adj
          / \     in contione
         /   \
        V'    \
       / \     Instr
      /   \    ---j
     V     \
  donarunt  \
            DO
           ---i

And finally:

       TopVP
       /  \
      /    \
    DO     TopVP
scribas_i  /  \
          /    \
       Instr  FocVP
   anulis_j   /  \
             /    \
          Adj     VP
  in contione_k  /  \
                /    \
               VP     Adj
              / \     ---k
             /   \
            V'    \
           / \     Instr
          /   \    ---j
         V     \
      donarunt  \
                DO
               ---i

This analysis shows that focus, the new information, is in contione, while scribas and anulis are topics, or old information.

Words that do not raise are tail information, so if we saw scribas anulis donarunt in contione, we would assume that anulis is new information, scribas is old information, and in contione is tail.

Examples

Movement of the subject

Here are some examples from De Bello Gallico where the subject, Caesar, is all over the place:

1. Caesar eius dextram prendit (1.20) Caesar took his right hand
2. Postridie eius diei Caesar... (1.51; 2.12) On the day after Caesar...
3. munitiones institutas Caesar parat perficere (1.83) Caesar prepares to finish the fortifications he had begun
4. Peragit concilium Caesar (6.4) Caesar finished the conference
5. paulo longius profrediendum existimabat Caesar (3.56) Caesar thought that he should advance a little further

Example (1) is the default order. Example (2) the authors explain by stating that subjects still count as initial when preceded by an adjoined adverbial.

In example (3), the direct object is now to the left of the subject. Caesar appears to the immediate left of the verb(s), indicating that Caesar is the focus, while the fortifications are known information.

Example (4) has the verb first, and this is actually a special case of ordering which will be dealt with in the next section.

Example (5) has Caesar as tail information: the sentence is not about Caesar (i.e. Caesar is not the topic, but rather the topic is what he thought), nor is Caesar new information, so he appears in the tail of the sentence.

Movement of direct and indirect objects

Here are two examples from Livy highlighting difference in order for direct and indirect objects:

1. senatus libertatem his civitatibus dedit (33.34.10) The Senate gave freedom to these states
2. praedam custodiendam... trecentis Cretensium dedit (42.65.5) Gave the spoils to three hundred Cretans to guard

In example (1), the direct object libertatem and the indirect object civitatibus appear in the normal order. Furthermore, libertatem is new information, while civitatibus is old information (as evidenced by the use of his).

However, in example (2), while the direct object praedam comes first and the indirect object trecentis comes second, as expected, in this sentence praedam is old information, while trecentis is new information.

The authors state that there is no easy way to determine pragmatic function based on ordering for direct and indirect objects. However, when the indirect object moves to the left of the direct object, or when the direct object moves to the left of the subject, the moved phrase becomes the topic of the sentence:

1. tantum modo aratoribus Metellus obsides non dedit (Cicero, In Verrem, 2.3.124) Metellus has just about given hostages to the farmers
2. urbanum veterem exercitum Fulvius consul C. Fulvio Flacco legato in Etruriam dedit ducendum (Livy, 27.8.12) The consul Fulvius gave the old city army to C. Fulvius Flaccus, his legate, to lead into Etruria

In example (1), the indirect object aratoribus has been raised to the left of the subject Metellus, making it the topic of the sentence.

Likewise, in example (2), the direct object exercitum has been moved to the left of the subject Fulvius, making it the topic of the sentence.

Sometimes, though, the direct object is actually intimately tied into the verb, as in negotium dare ut (to give the task of), so the direct object is not a true direct object, but rather the indefinite object mentioned above.

And then we have hyperbaton, the wrapping of one phrase inside another, as in aditum petentibus conveniundi non dabat (Nepos, 4.3.3) Would not give the opportunity of meeting to those petitioning. Here we have aditum conveniundi being the direct object, and petentibus, the indirect object, wrapped inside the direct object. The authors don't quite explain this immediately. There is an entire chapter devoted to hyperbaton.

Movement of the verb to the initial position

In sentences where the verb appears at the beginning, the verb has been raised right out of the verb phrase, above the topic. So, for example:

Regnavit Ancus annos quattuor et viginti (Livy, 1.35.1) Ancus reigned for 24 years

The default order would put the verb last, but it has raised to the first position. Ancus is still the topic of the sentence, and arguably annos is new information. Then why is regnavit not last? I have not been able to understand the authors' reasoning here.

Stacked adjectives

Purely pre- and postmodifiers

The authors show data which indicates that the order of adjectives when placed before the noun (premodifiers) is the same as that of English, and when placed after the noun (postmodifiers) is opposite that of English.

Thus, we have:

1. calida bubula urina (Columella, 6.11.1) With warm ox urine (not "with ox warm urine")
2. Fiscinas olearias Campanicas duas (Cato, 153.1) Two Campanian olive baskets (not "olive Campanian two baskets")

The authors point out a study which determines that the hierarchy for adjectives is as follows:

(Quality >) Size > Length > Height > Width > Weight > 
Temperature > Age > Shape > Color > Provenance > Material.

Mixed pre- and postmodifiers

There is also the not uncommon case of both pre- and postmodifiers, as in:

1. eruditissimos homines Asiaticos (De Orat, 3.43) The most learned men from Asia (or the most learned Asian men)

If these were purely postmodifiers, we would expect that since eruditissimos is a quality, and Asiaticos is provenance, the order should be homines Asiaticos eruditissimos, which is, as expected, the opposite order from English. However, here eruditissimos and not Asiaticos gets moved to the premodifier position. The reason for this is the phrase structure: ( (homines Asiaticos) eruditissimos) which means the inner phrase doesn't change when the outer phrase is shuffled around.

More important modifiers

If a modifier is considered more important than its noun, it can go first:

1. oliva novo fisco includitur et prelo subicitur (Columella, 12.51.1) The olives are enclosed in a new basket and put under the press.

Here, the fact that a basket is used is not important, but it is important that it should be a new basket.