As A&G note, “The i-stem was confused by even the Romans themselves.” There are a variety of variations with this stem present in all three grammatical genders, making it incredibly difficult to organize the data except in broad patterns and rote memorization. To that effective, I’m going to design a series of posts on the i-stem declension.
The basic neuter i-stem declension takes the stem (mari-) and converts the final i- to an e- in the nominative and accusative singular.
mare, maris (n.) sea
sedīle, sedīlis (n.) seat
(photo credit: Wiktionary).
In this basic output, the neuter i-stem is far more regular than its masculine and feminine counterparts:
- Nominative and accusative singular: -e
- Ablative singular: -ī
- Nominative and accusative plural: -ia
- Genitive plural: -ium
These are all regularized and there are no exceptions…except for the majority of nouns in the neuter i-stem declension, which don’t decline like this at all. Most neuter i-stem nouns have a consonantal base in -al or -ar, which is retained in all morphological forms. This causes only one change: these forms are animal, animalis, and not *animale, animalis. Everything else remains the same.
tribūnal, tribūnālis (n.) judge’s platform
There’s one feature that Wiktionary fails to capture in this chat. The -a- at the end of the stem is short in the nominative and accusative singular, but along everywhere else (see my lexical entry above). This is true of all i-stem nouns ending in –al or -ar.
calcar, calcāris (n.): spur
(Here they got the -a- right. Go figure.)
There you have it! The neuter i-stem declension. It’s fairly regular; it merely entails a large quantity of regular rules.
The Essential AG: 68-9
Latin for Addicts 
I’m confused. tribunal and calcar don’t appear to be parisyllabic, as the previous post said they should be.
I’m glad you’re following the i-Stem series I am doing, and I regret to report you that you will continue to be confused and disappointed by the extent to which this class of nouns defies true categorization.
Having read all the i-Stem material within A&G, I can promise you that there is no model. There are patterns, but no central force to the class.
I cannot promise you will not continue to see inconsistencies in this presentation. I can only assure you that I am parroting these inconsistencies from A&G, and thank you for being so attentive.