Lists and natural language

Hierarchical structure

From a natural language perspective, LOGO data structures are especially appropriate. We have already noted the basic parallelism between LOGO words and lists and the words and the sentences of a human language. As it happens, the parallelism goes deeper than that. The embedding of lists within lists which LOGO allows provides a direct counterpart to the hierarchical structure which we find in so many aspects of linguistic form.

Words and sentences

Natural language sentences are undoubtedly made up of words, but words in turn have quite often their own internal structure (aside from that of the mere sequence of letters). Cats, for instance, if we compare it with cat, can be decomposed into two constituents cat and s. Cats like milk thus has a structure which we can represent using a LOGO list as [[cat s] like milk].

And if we also want to claim that the words making up sentences are themselves packaged into groups or phrases, list structure deals with this notion very simply.

For example, the influential 19th century grammarian, Nesfield, proposed for the sentence The second master of the school has been teaching my sons Euclid since Thursday last an internal structure which we can directly represent with the rather nasty list:

[[The second master [of the school]] [[has been teaching] [[my sons] Euclid] [since Thursday last]]]

Lists and trees

List structures such as this are, in fact, no more than notational variants of the tree diagrams more commonly used by linguists (and others) to represent constituency relationships. Trees with branches are perhaps easier to follow than lists with multiple brackets, but the LOGO list for The second master of the school has been teaching my sons Euclid since Thursday last is none the less a direct linear encoding of the following diagram:

Indeed, aside from the absence of the labelling of constituents and our deliberate maintenance of the actual word order (which prevents a replication of Nesfield's two way primary split of the subject constituent), this diagram and the LOGO list which it reflects corresponds directly to Nesfield's own box diagram of the sentence structure:

As to the labelling, we could easily include that in our list by introducing the names of constituents as the first elements of the corresponding sub lists. Following his analysis exactly this time, the subject of Nesfield's sentence could in this way be represented, as

[subject [[nominative master][enlargement.of.nominative [[enlargement.1 the] [enlargement.2 second][enlargement.3 [of the school]]]]

Sounds and meanings

Words and sentences have, of course, not only a grammatical structure, they also have a sound structure and a meaning structure.


If you focus on sounds you can, for instance, think of words as composed of syllables. e.g. [mar ma lad(e)]. And although the idea is less familiar, even the individual elements of these syllables are regularly broken down into constituent features by linguists who work on the sounds of language. Part of the specification of the first element of marmalade for example, would show that it is formed by making a complete closure of the mouth with the lips while simultaneously allowing the exit of air through the nose with a lowered soft palate. We can easily reduce this verbose description to the list [stop labial nasal]. It is then not difficult to extend this idea to all of the successive sounds of marmalade and represent the word more fully as a list of three sub-lists (the syllables) each of which contains lists of segment features. Here is the first syllable:

[[[stop labial nasal] [vowel open]] . . .


Perhaps the most popular view of meaning identifies it with the information about words found in dictionary entries. In fact, even the simplest of dictionaries typically provides more than just information about the meaning of words and perhaps meanings are the least explicit parts of dictionary entries(footnote).

But whether you choose to accept the format of an existing dictionary as a model or more ambitiously aim to improve on the structure of meaning definitions, you will by now not be surprised to find that the organisation of a dictionary entry can be quite straightforwardly replicated in list structure.

You might for example consider representing the meaning of buy as a list containing, amongst other things

[[kind.of exchange][involves money][converse sell][agent human][goes.with [at place][from person][for price]] . . .].

Or you might directly represent the following entry from a dictionary of Marshallese:

baaj (bahaj)
Slang. From Engl. 2, 3.
Taken aback; halted; give up; pass, in card games; surrender.
Etal wot bwe ibaaj. I pass (in poker). Ibaaj . . . Ibaaj. I give up (in fighting).
by the list:
[[orthography baaj] [phonology bahaj] [status slang] [etymology english] [grammar.class [2 3]] [means [taken aback] [halted] [give up] [pass [in card games]] [surrender]] [examples [etal wot bwe ibaaj] [I pass [in poker]][ibaaj ibaaj] [I give up [in fighting]]]]

There is, of course, more to it . . .

As you proceed you will discover many other way in which data can be set up in LOGO for the handling of natural language. For the moment this brief overview of the notions should be sufficient to allow you to move on to the next page to consider the ways in which the user interacts with a LOGO system in order to instruct it to perform whatever tasks are required.

[footnote] It is also worth remembering here that alongside word meaning - which dictionaries aim to handle - you may also want to talk about other types of meaning such as sentence meaning. 'He has been teaching my sons' does not mean the same thing as 'Has he been teaching my sons?' although the words are the same.BACK

Ron Brasington
Department of Linguistic Science
The University of Reading